Remove CodeBLEU dependency and use tree-sitter directly

Merge main branch
Don't try to send the new events in the UI (#7277 )
2026-04-29 03:00:45 -04:00 · 2025-03-17 16:45:19 +00:00 · 2025-03-17 16:43:36 +00:00 · 2025-03-17 14:50:22 +01:00 · 2025-03-17 21:15:01 +08:00 · 2025-03-17 05:50:05 +00:00
119 changed files with 14268 additions and 1132 deletions
@@ -12,7 +12,7 @@
  <a href="https://codecov.io/github/All-Hands-AI/OpenHands?branch=main"><img alt="CodeCov" src="https://img.shields.io/codecov/c/github/All-Hands-AI/OpenHands?style=for-the-badge&color=blue"></a>
  <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/LICENSE"><img src="https://img.shields.io/github/license/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" alt="MIT License"></a>
  <br/>
-  <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community"></a>
+  <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community"></a>
  <a href="https://discord.gg/ESHStjSjD4"><img src="https://img.shields.io/badge/Discord-Join%20Us-purple?logo=discord&logoColor=white&style=for-the-badge" alt="Join our Discord community"></a>
  <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/CREDITS.md"><img src="https://img.shields.io/badge/Project-Credits-blue?style=for-the-badge&color=FFE165&logo=github&logoColor=white" alt="Credits"></a>
  <br/>
@@ -96,7 +96,7 @@ troubleshooting resources, and advanced configuration options.
 OpenHands is a community-driven project, and we welcome contributions from everyone. We do most of our communication
 through Slack, so this is the best place to start, but we also are happy to have you contact us on Discord or Github:

- [Join our Slack workspace](https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg) - Here we talk about research, architecture, and future development.
+- [Join our Slack workspace](https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw) - Here we talk about research, architecture, and future development.
 - [Join our Discord server](https://discord.gg/ESHStjSjD4) - This is a community-run server for general discussion, questions, and feedback.
 - [Read or post Github Issues](https://github.com/All-Hands-AI/OpenHands/issues) - Check out the issues we're working on, or add your own ideas.

@@ -42,7 +42,7 @@ Explorez le code source d'OpenHands sur [GitHub](https://github.com/All-Hands-AI
  />
 </a>
 <br></br>
-<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg">
+<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw">
  <img
    src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge"
    alt="Join our Slack community"
@@ -42,7 +42,7 @@ OpenHands 是一个**自主 AI 软件工程师**，能够执行复杂的工程
  />
 </a>
 <br></br>
-<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg">
+<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw">
  <img
    src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge"
    alt="Join our Slack community"
@@ -8,7 +8,7 @@ function CustomFooter() {
    <footer className="custom-footer">
      <div className="footer-content">
        <div className="footer-icons">
-          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg" target="_blank" rel="noopener noreferrer">
+          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw" target="_blank" rel="noopener noreferrer">
            <FaSlack />
          </a>
          <a href="https://discord.gg/ESHStjSjD4" target="_blank" rel="noopener noreferrer">
@@ -46,7 +46,7 @@ export function HomepageHeader() {
          <a href="https://codecov.io/github/All-Hands-AI/OpenHands?branch=main"><img alt="CodeCov" src="https://img.shields.io/codecov/c/github/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" /></a>
          <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/LICENSE"><img src="https://img.shields.io/github/license/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" alt="MIT License" /></a>
          <br/>
-          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community" /></a>
+          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ngejmfw6-9gW4APWOC9XUp1n~SiQ6iw"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community" /></a>
          <a href="https://discord.gg/ESHStjSjD4"><img src="https://img.shields.io/badge/Discord-Join%20Us-purple?logo=discord&logoColor=white&style=for-the-badge" alt="Join our Discord community" /></a>
          <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/CREDITS.md"><img src="https://img.shields.io/badge/Project-Credits-blue?style=for-the-badge&color=FFE165&logo=github&logoColor=white" alt="Credits" /></a>
          <br/>
@@ -1,3 +1,4 @@
+import copy
 import json
 import os
 import subprocess
@@ -175,6 +176,11 @@ def process_instance(
        logger.warning(
            f'This is the {runtime_failure_count + 1}th attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
        )
+    metadata = copy.deepcopy(metadata)
+    metadata.details['runtime_failure_count'] = runtime_failure_count
+    metadata.details['remote_runtime_resource_factor'] = (
+        config.sandbox.remote_runtime_resource_factor
+    )

    try:
        runtime = create_runtime(config)
@@ -296,14 +302,20 @@ def process_instance(
                        with open(test_output_path, 'w') as f:
                            f.write(test_output)
                        try:
+                            extra_kwargs = {}
+                            if 'SWE-Gym' in metadata.dataset:
+                                # SWE-Gym uses a different version of the package, hence a different eval report argument
+                                extra_kwargs['log_path'] = test_output_path
+                            else:
+                                extra_kwargs['test_log_path'] = test_output_path
                            _report = conditional_imports.get_eval_report(
                                test_spec=test_spec,
                                prediction={
                                    'model_patch': model_patch,
                                    'instance_id': instance_id,
                                },
-                                test_log_path=test_output_path,
                                include_tests_status=True,
+                                **extra_kwargs,
                            )
                            report = _report[instance_id]
                            logger.info(
@@ -463,6 +475,7 @@ if __name__ == '__main__':
            .decode('utf-8')
            .strip(),  # Current commit
            dataset=args.dataset,  # Dataset name from args
+            details={},
        )

    # The evaluation harness constrains the signature of `process_instance_func` but we need to
@@ -23,7 +23,7 @@ def get_resource_mapping(dataset_name: str) -> dict[str, float]:
    if dataset_name not in _global_resource_mapping:
        file_path = os.path.join(CUR_DIR, f'{dataset_name}.json')
        if not os.path.exists(file_path):
-            logger.warning(f'Resource mapping for {dataset_name} not found.')
+            logger.info(f'Resource mapping for {dataset_name} not found.')
            return None

        with open(file_path, 'r') as f:
@@ -1,4 +1,5 @@
 import asyncio
+import copy
 import json
 import os
 import tempfile
@@ -149,7 +150,8 @@ def get_config(
 ) -> AppConfig:
    # We use a different instance image for the each instance of swe-bench eval
    use_official_image = bool(
-        'verified' in metadata.dataset.lower() or 'lite' in metadata.dataset.lower()
+        ('verified' in metadata.dataset.lower() or 'lite' in metadata.dataset.lower())
+        and 'swe-gym' not in metadata.dataset.lower()
    )
    base_container_image = get_instance_docker_image(
        instance['instance_id'], use_official_image
@@ -475,6 +477,13 @@ def process_instance(
        logger.warning(
            f'This is the {runtime_failure_count + 1}th attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
        )
+
+    metadata = copy.deepcopy(metadata)
+    metadata.details['runtime_failure_count'] = runtime_failure_count
+    metadata.details['remote_runtime_resource_factor'] = (
+        config.sandbox.remote_runtime_resource_factor
+    )
+
    runtime = create_runtime(config)
    call_async_from_sync(runtime.connect)

@@ -560,20 +569,6 @@ def filter_dataset(dataset: pd.DataFrame, filter_column: str) -> pd.DataFrame:
    return dataset


-# A list of instances that are known to be tricky to infer
-# (will cause runtime failure even with resource factor = 8)
-SWEGYM_EXCLUDE_IDS = [
-    'dask__dask-10422',
-    'pandas-dev__pandas-50548',
-    'pandas-dev__pandas-53672',
-    'pandas-dev__pandas-54174',
-    'pandas-dev__pandas-55518',
-    'pandas-dev__pandas-58383',
-    'pydata__xarray-6721',
-    'pytest-dev__pytest-10081',
-    'pytest-dev__pytest-7236',
-]
-
 if __name__ == '__main__':
    parser = get_parser()
    parser.add_argument(
@@ -598,11 +593,20 @@ if __name__ == '__main__':
        f'Loaded dataset {args.dataset} with split {args.split}: {len(swe_bench_tests)} tasks'
    )
    if 'SWE-Gym' in args.dataset:
-        swe_bench_tests = swe_bench_tests[
-            ~swe_bench_tests['instance_id'].isin(SWEGYM_EXCLUDE_IDS)
-        ]
+        with open(
+            os.path.join(
+                os.path.dirname(os.path.abspath(__file__)),
+                'split',
+                'swegym_verified_instances.json',
+            ),
+            'r',
+        ) as f:
+            swegym_verified_instances = json.load(f)
+            swe_bench_tests = swe_bench_tests[
+                swe_bench_tests['instance_id'].isin(swegym_verified_instances)
+            ]
        logger.info(
-            f'{len(swe_bench_tests)} tasks left after excluding SWE-Gym excluded tasks'
+            f'{len(swe_bench_tests)} tasks left after filtering for SWE-Gym verified instances'
        )

    llm_config = None
@@ -9,7 +9,7 @@ parser.add_argument(
    '--dataset_name',
    type=str,
    help='Name of the dataset to download',
-    default='princeton-nlp/SWE-bench_Lite',
+    default='princeton-nlp/SWE-bench_Verified',
 )
 parser.add_argument('--split', type=str, help='Split to download', default='test')
 args = parser.parse_args()
@@ -20,7 +20,12 @@ print(
    f'Downloading gold patches from {args.dataset_name} (split: {args.split}) to {output_filepath}'
 )
 patches = [
-    {'instance_id': row['instance_id'], 'model_patch': row['patch']} for row in dataset
+    {
+        'instance_id': row['instance_id'],
+        'model_patch': row['patch'],
+        'model_name_or_path': 'gold',
+    }
+    for row in dataset
 ]
 print(f'{len(patches)} gold patches loaded')
 pd.DataFrame(patches).to_json(output_filepath, lines=True, orient='records')
@@ -0,0 +1,12 @@
+codamosa_ids = ['pydata__xarray-4750-16496', 'pydata__xarray-3239-16458', 'pydata__xarray-4966-16515', 'pydata__xarray-3302-16459', 'pydata__xarray-5126-16518', 'pydata__xarray-4994-16516', 'pydata__xarray-3905-16478', 'pydata__xarray-4182-16484', 'pydata__xarray-5131-16520', 'pydata__xarray-5662-16532', 'pydata__xarray-3364-16461', 'pydata__xarray-5731-16534', 'pydata__xarray-3239-16457', 'pydata__xarray-7203-16577', 'pydata__xarray-3156-16454', 'pydata__xarray-5126-16519', 'pydata__xarray-5365-16529', 'pydata__xarray-4629-16492', 'pydata__xarray-4248-16486', 'pydata__xarray-4339-16487', 'pydata__xarray-3151-16453', 'pydata__xarray-3114-16452', 'pydata__xarray-5033-16517', 'pydata__xarray-4802-16505', 'pydata__xarray-5455-16530', 'pydata__xarray-6400-16539', 'pydata__xarray-3239-16456', 'pydata__xarray-4419-16488']
+
+pynguin_ids = ['pydata__xarray-6548-16541', 'pydata__xarray-7003-16557', 'pydata__xarray-3114-16452', 'pydata__xarray-4339-16487', 'pydata__xarray-6889-16549', 'pydata__xarray-3239-16458', 'pydata__xarray-3364-16461', 'pydata__xarray-3239-16457', 'pydata__xarray-5365-16529', 'pydata__xarray-5131-16520', 'pydata__xarray-7229-16578', 'pydata__xarray-6461-16540', 'pydata__xarray-4419-16488', 'pydata__xarray-7147-16571', 'pydata__xarray-3151-16453', 'pydata__xarray-4966-16515', 'pydata__xarray-4629-16492', 'pydata__xarray-3239-16456', 'pydata__xarray-7400-16582', 'pydata__xarray-4994-16516', 'pydata__xarray-3302-16459', 'pydata__xarray-6601-16544', 'pydata__xarray-6882-16548', 'pydata__xarray-6135-16535', 'pydata__xarray-7393-16581', 'pydata__xarray-5731-16534', 'pydata__xarray-7203-16577']
+
+ids = ['pydata__xarray-3114-16452', 'pydata__xarray-3151-16453', 'pydata__xarray-3156-16454', 'pydata__xarray-3239-16456', 'pydata__xarray-3239-16457', 'pydata__xarray-3239-16458', 'pydata__xarray-3302-16459', 'pydata__xarray-3364-16461', 'pydata__xarray-3677-16471', 'pydata__xarray-3905-16478', 'pydata__xarray-4182-16484', 'pydata__xarray-4248-16486', 'pydata__xarray-4339-16487', 'pydata__xarray-4419-16488', 'pydata__xarray-4629-16492', 'pydata__xarray-4750-16496', 'pydata__xarray-4802-16505', 'pydata__xarray-4966-16515', 'pydata__xarray-4994-16516', 'pydata__xarray-5033-16517', 'pydata__xarray-5126-16518', 'pydata__xarray-5126-16519', 'pydata__xarray-5131-16520', 'pydata__xarray-5365-16529', 'pydata__xarray-5455-16530', 'pydata__xarray-5662-16532', 'pydata__xarray-5731-16534', 'pydata__xarray-6135-16535', 'pydata__xarray-6135-16536', 'pydata__xarray-6386-16537', 'pydata__xarray-6394-16538', 'pydata__xarray-6400-16539', 'pydata__xarray-6461-16540', 'pydata__xarray-6548-16541', 'pydata__xarray-6599-16543', 'pydata__xarray-6601-16544', 'pydata__xarray-6882-16548', 'pydata__xarray-6889-16549', 'pydata__xarray-7003-16557', 'pydata__xarray-7147-16571', 'pydata__xarray-7150-16572', 'pydata__xarray-7203-16577', 'pydata__xarray-7229-16578', 'pydata__xarray-7393-16581', 'pydata__xarray-7400-16582']
+
+
+Command eval (our approach):
+poetry run ./evaluation/benchmarks/testgeneval/scripts/eval_infer_remote.sh evaluation/evaluation_outputs/outputs/kjain14__testgeneval-test/CodeActAgent/gpt-4o_maxiter_25_N_v0.20.0-no-hint-run_1/output.jsonl 10 kjain14/testgeneval test true
+
+Command run (our approach):
+./evaluation/benchmarks/testgeneval/scripts/run_infer.sh llm.eval_gpt HEAD CodeActAgent -1 25 10 kjain14/testgeneval test 1 ../TestGenEval/results/testgeneval/preds/gpt-4o-2024-08-06__testgeneval__0.2__test.jsonl
@@ -0,0 +1,80 @@
+# TestGenEval Benchmark Evaluation
+
+This folder contains the evaluation harness for the TestGenEval benchmark, which is based on the original TestGenEval benchmark ([paper](https://arxiv.org/abs/2410.00752)). TestGenEval is designed to evaluate the ability of language models to generate unit tests for given Python functions.
+
+## Setup Environment and LLM Configuration
+
+1. Follow the instructions [here](../../README.md#setup) to set up your local development environment and configure your LLM.
+
+2. Install the TestGenEval dependencies:
+```bash
+poetry install --with testgeneval
+```
+
+## Run Inference
+
+To generate tests using your model, run the following command:
+
+```bash
+./evaluation/benchmarks/testgeneval/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit] [max_iter] [num_workers] [dataset] [dataset_split]
+
+# Example
+./evaluation/benchmarks/testgeneval/scripts/run_infer.sh llm.eval_gpt4_1106_preview HEAD CodeActAgent 100 30 1 kjain14/testgenevallite test
+```
+
+Parameters:
+- `model_config`: The config group name for your LLM settings (e.g., `eval_gpt4_1106_preview`)
+- `git-version`: The git commit hash or release tag of OpenHands to evaluate (e.g., `HEAD` or `0.6.2`)
+- `agent`: The name of the agent for benchmarks (default: `CodeActAgent`)
+- `eval_limit`: Limit the evaluation to the first N instances (optional)
+- `max_iter`: Maximum number of iterations for the agent to run (default: 30)
+- `num_workers`: Number of parallel workers for evaluation (default: 1)
+- `dataset`: HuggingFace dataset name (default: `kjain14/testgenevallite`)
+- `dataset_split`: Dataset split to use (default: `test`)
+
+After running the inference, you will obtain an `output.jsonl` file (by default saved to `evaluation/evaluation_outputs`).
+
+## Evaluate Generated Tests
+
+To evaluate the generated tests, use the `eval_infer.sh` script:
+
+```bash
+./evaluation/benchmarks/testgeneval/scripts/eval_infer.sh $YOUR_OUTPUT_JSONL [instance_id] [dataset_name] [split] [num_workers] [skip_mutation]
+
+# Example
+./evaluation/benchmarks/testgeneval/scripts/eval_infer.sh evaluation/evaluation_outputs/outputs/testgeneval/CodeActAgent/gpt-4-1106-preview_maxiter_50_N_v1.0/output.jsonl
+```
+
+Optional arguments:
+- `instance_id`: Evaluate a single instance (optional)
+- `dataset_name`: Name of the dataset to use (default: `kjain14/testgenevallite`)
+- `split`: Dataset split to use (default: `test`)
+- `num_workers`: Number of workers for running docker (default: 1)
+- `skip_mutation`: Skip mutation testing (enter `true` if desired)
+
+The evaluation results will be saved to `evaluation/evaluation_outputs/outputs/testgeneval/CodeActAgent/gpt-4-1106-preview_maxiter_50_N_v1.0/` with `output.testgeneval.jsonl` containing the metrics.
+
+## Metrics
+
+The TestGenEval benchmark evaluates generated tests based on the following metrics:
+
+1. Correctness: Measures if the generated tests are syntactically correct and run without errors.
+2. Coverage: Assesses the code coverage achieved by the generated tests.
+3. Mutation Score: Evaluates the effectiveness of the tests in detecting intentionally introduced bugs (mutations).
+4. Readability: Analyzes the readability of the generated tests using various metrics.
+
+## Submit Your Evaluation Results
+
+To contribute your evaluation results:
+
+1. Fork [our HuggingFace evaluation outputs](https://huggingface.co/spaces/OpenHands/evaluation).
+2. Add your results to the forked repository.
+3. Submit a Pull Request with your evaluation results following the guide [here](https://huggingface.co/docs/hub/en/repositories-pull-requests-discussions#pull-requests-and-discussions).
+
+## Additional Resources
+
+- [TestGenEval Paper](https://arxiv.org/abs/2410.00752)
+- [OpenHands Documentation](https://github.com/All-Hands-AI/OpenHands)
+- [HuggingFace Datasets](https://huggingface.co/datasets)
+
+For any questions or issues, please open an issue in the [OpenHands repository](https://github.com/All-Hands-AI/OpenHands/issues).
@@ -0,0 +1,356 @@
+import math
+import os
+from pathlib import Path
+
+from tree_sitter import Language, Parser
+
+
+def total_byte_entropy_stats(python_code):
+    # Count the occurrence of each byte (character for simplicity)
+    byte_counts = {}
+    for byte in python_code.encode('utf-8'):
+        byte_counts[byte] = byte_counts.get(byte, 0) + 1
+
+    total_bytes = sum(byte_counts.values())
+    entropy = -sum(
+        (count / total_bytes) * math.log2(count / total_bytes)
+        for count in byte_counts.values()
+    )
+
+    return {'total_byte_entropy': entropy}
+
+
+def average_nulls_stats(tree, num_lines):
+    total_nulls = 0
+    nulls_per_line = {}  # Dictionary to count nulls per line
+
+    def traverse(node):
+        nonlocal total_nulls
+        if node.type == 'null_literal':
+            total_nulls += 1
+            line_number = node.start_point[0]  # Get line number
+            if line_number in nulls_per_line:
+                nulls_per_line[line_number] += 1
+            else:
+                nulls_per_line[line_number] = 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+
+    # Calculate average nulls per line
+    avg_nulls = total_nulls / num_lines if num_lines > 0 else 0
+
+    # Calculate max nulls on any line
+    max_nulls_on_any_line = max(nulls_per_line.values()) if nulls_per_line else 0
+
+    return {
+        'avg_nulls': avg_nulls,
+        'total_nulls': total_nulls,
+        'max_nulls': max_nulls_on_any_line,
+        'has_nulls': 1 if total_nulls > 0 else 0,
+    }
+
+
+def arithmetic_operations_stats(tree, num_lines):
+    # Dictionary to hold counts of each arithmetic operation
+    op_counts = {'+': 0, '-': 0, '*': 0, '/': 0, '%': 0}
+    total_ops = 0
+
+    # Function to traverse the AST and update operation counts
+    def traverse(node):
+        nonlocal total_ops
+        if node.type == 'binary_expression' or node.type == 'update_expression':
+            for child in node.children:
+                if child.type == 'operator':
+                    op = child.text.decode('utf8')
+                    if op in op_counts:
+                        op_counts[op] += 1
+                        total_ops += 1
+        else:
+            for child in node.children:
+                traverse(child)
+
+    traverse(tree.root_node)
+
+    return {
+        'total_arithmetic_operations': total_ops,
+        'avg_arithmetic_operations': total_ops / num_lines,
+    }
+
+
+def numbers_floats_stats(tree, num_lines):
+    total_numbers = 0
+    total_floats = 0
+
+    def traverse(node):
+        nonlocal total_numbers, total_floats
+        if node.type in ['integer_literal', 'decimal_literal']:
+            total_numbers += 1
+            if (
+                '.' in node.text.decode('utf8')
+                or 'e' in node.text.decode('utf8').lower()
+            ):
+                total_floats += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    return {'total_numbers': total_numbers, 'total_floats': total_floats}
+
+
+def code_stats(python_code):
+    lines = python_code.strip().split('\n')
+    total_line_length = sum(len(line) for line in lines)
+    max_line_length = max(len(line) for line in lines)
+    return {
+        'total_line_length': total_line_length,
+        'max_line_length': max_line_length,
+        'avg_characters': total_line_length / len(lines),
+    }
+
+
+def assertions_stats(tree, num_lines):
+    total_assertions = 0
+
+    def traverse(node):
+        nonlocal total_assertions
+        if node.type == 'assert_statement':
+            total_assertions += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    return {
+        'total_assertions': total_assertions,
+        'total_has_assertions': 1 if total_assertions > 0 else 0,
+    }
+
+
+def class_instances_stats(tree, num_lines):
+    total_class_instances = 0
+
+    def traverse(node):
+        nonlocal total_class_instances
+        if node.type == 'object_creation_expression':
+            total_class_instances += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    return {'total_class_instances': total_class_instances}
+
+
+def has_execeptions(tree, num_lines):
+    total_has_exceptions = 0
+
+    def traverse(node):
+        nonlocal total_has_exceptions
+        if node.type == 'try_statement':
+            total_has_exceptions += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    return {'total_has_exceptions': 1 if total_has_exceptions > 0 else 0}
+
+
+def distinct_methods_stats(tree, num_lines):
+    method_names = set()
+    total_nodes = 0
+
+    def traverse(node):
+        nonlocal total_nodes
+        if node.type == 'method_declaration':
+            for child in node.children:
+                if child.type == 'identifier':
+                    method_names.add(child.text.decode('utf8'))
+                    break
+        total_nodes += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    total_distinct_methods = len(method_names)
+    total_method_ratio = (
+        total_distinct_methods / (total_nodes - total_distinct_methods)
+        if total_nodes > total_distinct_methods
+        else 0
+    )
+
+    return {
+        'total_distinct_methods': total_distinct_methods,
+        'total_method_ratio': total_method_ratio,
+    }
+
+
+def loops_stats(tree, num_lines):
+    """
+    Calculate the average number of loops.
+    """
+    total_loops = 0
+
+    def traverse(node):
+        nonlocal total_loops
+        if node.type in ['for_statement', 'while_statement', 'do_statement']:
+            total_loops += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    avg_loops = total_loops / num_lines
+    return {'avg_loops': avg_loops}
+
+
+def branches_stats(tree, num_lines):
+    """
+    Calculate the average number of branches (conditional statements).
+    """
+    total_branches = 0
+
+    def traverse(node):
+        nonlocal total_branches
+        if node.type in ['if_statement', 'switch_statement']:
+            total_branches += 1
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+    # Assuming each branch is its own, this might need refinement based on definition
+    avg_branches = total_branches / num_lines
+    return {'avg_branches': avg_branches}
+
+
+def string_stats(tree, num_lines):
+    string_literals = []
+
+    # Function to traverse the AST and collect string literals
+    def traverse(node):
+        if node.type == 'string_literal':
+            # Extracting the string literal, excluding the quotation marks
+            literal_text = node.text.decode('utf8')[1:-1]
+            string_literals.append(literal_text)
+        for child in node.children:
+            traverse(child)
+
+    traverse(tree.root_node)
+
+    # Calculate the average string length
+    total_length = sum(len(s) for s in string_literals)
+    avg_length = total_length / num_lines
+    return {'avg_str_length': avg_length}
+
+
+def identifier_stats(tree, num_lines):
+    root_node = tree.root_node
+    identifier_counts = {}  # Dictionary to count occurrences of each identifier
+    total_nodes = 0  # Counter for all nodes
+
+    # Function to recursively count identifiers and all nodes, gathering their stats
+    def count(node):
+        nonlocal identifier_counts, total_nodes
+        iden_count = 0
+        max_length = 0
+        total_nodes += 1  # Increment total nodes for every node visited
+        if node.type == 'identifier':
+            identifier = node.text.decode('utf8')  # Assuming UTF-8 encoding
+            iden_count += 1
+            identifier_counts[identifier] = identifier_counts.get(identifier, 0) + 1
+            iden_length = len(identifier)
+            if iden_length > max_length:
+                max_length = iden_length
+        for child in node.children:
+            child_count, child_max_length = count(child)
+            iden_count += child_count
+            if child_max_length > max_length:
+                max_length = child_max_length
+        return iden_count, max_length
+
+    total_identifiers, max_identifier_length = count(root_node)
+    total_unique_identifiers = len(identifier_counts)
+    total_identifier_length = sum(len(k) * v for k, v in identifier_counts.items())
+    avg_identifier_length = total_identifier_length / num_lines
+
+    # Calculate the identifier ratio as total identifiers over total nodes
+    identifier_ratio = total_identifiers / total_nodes if total_nodes > 0 else 0
+
+    return {
+        'total_identifiers': total_identifiers,
+        'total_identifier_length': total_identifier_length,
+        'max_identifier_length': max_identifier_length,
+        'avg_identifier_length': avg_identifier_length,
+        'total_unique_identifiers': total_unique_identifiers,
+        'total_identifier_ratio': identifier_ratio,  # Include the new ratio in the returned dictionary
+        'total_nodes': total_nodes,  # Include total node count for reference or further calculations
+    }
+
+
+def compute_regression(results):
+    components = {
+        'total_line_length': -0.0001,
+        'max_line_length': -0.0021,
+        'total_identifiers': 0.0076,
+        'total_identifier_length': -0.0004,
+        'max_identifier_length': -0.0067,
+        'avg_identifier_length': -0.005,
+        'avg_arithmetic_operations': 0.0225,
+        'avg_branches': 0.9886,
+        'avg_loops': 0.1572,
+        'total_assertions': 0.0119,
+        'total_has_assertions': -0.0147,
+        'avg_characters': 0.1242,
+        'total_class_instances': -0.043,
+        'total_distinct_methods': -0.0127,
+        'avg_str_length': 0.0026,
+        'total_has_exceptions': 0.1206,
+        'total_unique_identifiers': -0.019,
+        'max_nulls': -0.0712,
+        'total_numbers': -0.0078,
+        'avg_nulls': 0.1444,
+        'total_identifier_ratio': 0.334,
+        'total_method_ratio': 0.0406,
+        'total_floats': -0.0174,
+        'total_byte_entropy': -0.3917,
+    }
+    test_score = 0
+
+    for component in components:
+        test_score += components[component] * results[component]
+
+    test_score += 5.7501
+    return test_score
+
+
+def compute_readability(python_code):
+    parser = Parser()
+    this_dir = Path(os.path.dirname(os.path.realpath(__file__)))
+    parser.set_language(Language.build_library(
+        # Store the library in the `build` directory
+        this_dir / "build" / "my-languages.so",
+        # Include one or more languages
+        [
+            this_dir / "tree-sitter-python"
+        ]
+    ).get_language('python'))
+
+    results = code_stats(python_code)
+
+    num_lines = len(python_code.strip().split('\n'))
+    results.update(total_byte_entropy_stats(python_code))
+
+    tree = parser.parse(bytes(python_code, 'utf8'))
+
+    results.update(identifier_stats(tree, num_lines))
+    results.update(loops_stats(tree, num_lines))
+    results.update(branches_stats(tree, num_lines))
+    results.update(distinct_methods_stats(tree, num_lines))
+    results.update(has_execeptions(tree, num_lines))
+    results.update(class_instances_stats(tree, num_lines))
+    results.update(assertions_stats(tree, num_lines))
+    results.update(numbers_floats_stats(tree, num_lines))
+    results.update(average_nulls_stats(tree, num_lines))
+    results.update(arithmetic_operations_stats(tree, num_lines))
+    results.update(string_stats(tree, num_lines))
+
+    score = compute_regression(results)
+    return score
@@ -0,0 +1,633 @@
+import os
+import tempfile
+import time
+from functools import partial
+
+import pandas as pd
+from report_utils import (
+    check_coverage,
+    check_mutation,
+    count_methods,
+    get_lines_of_code,
+)
+
+from evaluation.benchmarks.testgeneval.compute_readability import compute_readability
+from evaluation.benchmarks.testgeneval.constants import (
+    COVERAGE_PREFIX,
+    MUTATION_BUFFER,
+    MUTATION_TEMPLATE,
+    MUTATION_TIMEOUT,
+    TESTS_SUFFIX,
+)
+from evaluation.benchmarks.testgeneval.metrics import (
+    bleu,
+    code_bleu,
+    edit_sim,
+    exact_match,
+    rouge_l,
+)
+from evaluation.benchmarks.testgeneval.pygments_utils import tokenize_code
+from evaluation.benchmarks.testgeneval.run_infer import get_instance_docker_image
+from evaluation.benchmarks.testgeneval.test_filter import filter_tests
+from evaluation.benchmarks.testgeneval.test_spec import (
+    TestGenEvalInstance,
+    TestSpec,
+    make_test_spec,
+)
+from evaluation.benchmarks.testgeneval.utils import load_testgeneval_dataset
+from evaluation.utils.shared import (
+    EvalMetadata,
+    EvalOutput,
+    prepare_dataset,
+    reset_logger_for_multiprocessing,
+    run_evaluation,
+)
+from openhands.core.config import AppConfig, SandboxConfig, get_parser
+from openhands.core.logger import openhands_logger as logger
+from openhands.core.main import create_runtime
+from openhands.events.action import CmdRunAction
+from openhands.events.observation import CmdOutputObservation
+from openhands.utils.async_utils import call_async_from_sync
+
+DOCKER_IMAGE_PREFIX = os.environ.get('EVAL_DOCKER_IMAGE_PREFIX', 'docker.io/kdjain/')
+logger.info(f'Using docker image prefix: {DOCKER_IMAGE_PREFIX}')
+
+
+def get_config(instance: pd.Series) -> AppConfig:
+    base_container_image = get_instance_docker_image(instance['instance_id_swebench'])
+    assert (
+        base_container_image
+    ), f"Invalid container image for instance {instance['instance_id_swebench']}."
+    logger.info(f'Using instance container image: {base_container_image}.')
+    return AppConfig(
+        run_as_openhands=False,
+        runtime=os.environ.get('RUNTIME', 'eventstream'),
+        sandbox=SandboxConfig(
+            base_container_image=base_container_image,
+            use_host_network=False,
+            timeout=1800,
+            api_key=os.environ.get('ALLHANDS_API_KEY'),
+            remote_runtime_api_url=os.environ.get(
+                'SANDBOX_REMOTE_RUNTIME_API_URL', 'http://localhost:8000'
+            ),
+        ),
+        workspace_base=None,
+        workspace_mount_path=None,
+    )
+
+
+def compute_lexical_metrics(pred_suite, gold_suite):
+    pred_loc = get_lines_of_code(pred_suite)
+    gold_loc = get_lines_of_code(gold_suite)
+    pred_methods = count_methods(pred_suite)
+    gold_methods = count_methods(gold_suite)
+    readability_pred = compute_readability(pred_suite)
+    readability_gold = compute_readability(gold_suite)
+
+    preds = tokenize_code(pred_suite)
+    golds = tokenize_code(gold_suite)
+
+    return {
+        'pred_loc': pred_loc,
+        'gold_loc': gold_loc,
+        'pred_readability': readability_pred,
+        'gold_readability': readability_gold,
+        'pred_methods': pred_methods,
+        'gold_methods': gold_methods,
+        'code_bleu': code_bleu(preds, golds, 'Python3'),
+        'bleu': bleu(preds, golds),
+        'xmatch': exact_match(preds, golds),
+        'edit_sim': edit_sim(preds, golds),
+        'rouge_f': rouge_l(golds, preds)['f'],
+        'rouge_p': rouge_l(golds, preds)['p'],
+        'rouge_r': rouge_l(golds, preds)['r'],
+    }
+
+
+def run_command(runtime, command, timeout=600):
+    action = CmdRunAction(command=command)
+    action.set_hard_timeout(timeout)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert obs.exit_code == 0
+    return obs
+
+
+def run_tests(runtime, instance, test_script, log_file='/tmp/test_output.log'):
+    action = CmdRunAction(command=f'bash {test_script} > {log_file} 2>&1 & echo $!')
+    action.set_hard_timeout(60)
+    obs = runtime.run_action(action)
+
+    assert isinstance(obs, CmdOutputObservation), 'Failed to start test script.'
+    pid = obs.content.split()[-1].strip()
+    logger.info(f'[{instance.instance_id}] Test process started with PID: {pid}')
+
+    start_time = time.time()
+    timeout = 1800
+    while True:
+        elapsed_time = time.time() - start_time
+        if elapsed_time > timeout:
+            logger.info(f'[{instance.instance_id}] Test process timed out.')
+            instance['test_result']['report']['test_timeout'] = True
+            break
+
+        check_action = CmdRunAction(command=f'ps -p {pid} > /dev/null; echo $?')
+        check_obs = runtime.run_action(check_action)
+        if (
+            isinstance(check_obs, CmdOutputObservation)
+            and len(check_obs.content.split()) > 0
+            and check_obs.content.split()[-1].strip() == '1'
+        ):
+            logger.info(f'[{instance.instance_id}] Test process completed.')
+            break
+        time.sleep(30)
+
+    test_action = CmdRunAction(command=f'cat {log_file}')
+    test_action.set_hard_timeout(300)
+    test_obs = runtime.run_action(test_action)
+    assert isinstance(test_obs, CmdOutputObservation), 'Failed to retrieve test output.'
+    return test_obs.exit_code, test_obs.content, elapsed_time
+
+
+def run_mutation_testing(
+    runtime, instance, mutation_script, log_file='/tmp/mutation_output.log'
+):
+    action = CmdRunAction(command=f'bash {mutation_script} > {log_file} 2>&1 & echo $!')
+    action.set_hard_timeout(60)
+    obs = runtime.run_action(action)
+
+    assert isinstance(obs, CmdOutputObservation), 'Failed to start test script.'
+    pid = obs.content.split()[-1].strip()
+    logger.info(f'[{instance.instance_id}] Mutation process started with PID: {pid}')
+
+    start_time = time.time()
+    timeout = 4000
+    while True:
+        elapsed_time = time.time() - start_time
+        if elapsed_time > timeout:
+            logger.info(f'[{instance.instance_id}] Mutation process timed out.')
+            instance['test_result']['report']['mutation_timeout'] = True
+            break
+
+        check_action = CmdRunAction(command=f'ps -p {pid} > /dev/null; echo $?')
+        check_obs = runtime.run_action(check_action)
+        if (
+            isinstance(check_obs, CmdOutputObservation)
+            and len(check_obs.content.split()) > 0
+            and check_obs.content.split()[-1].strip() == '1'
+        ):
+            logger.info(f'[{instance.instance_id}] Mutation process completed.')
+            break
+        time.sleep(30)
+
+    assert isinstance(obs, CmdOutputObservation), 'Failed to run mutation script.'
+    mutation_action = CmdRunAction(command=f'cat {log_file}')
+    mutation_action.set_hard_timeout(300)
+    mutation_obs = runtime.run_action(mutation_action)
+    assert isinstance(
+        mutation_obs, CmdOutputObservation
+    ), 'Failed to retrieve mutation output.'
+    return mutation_obs.exit_code, mutation_obs.content
+
+
+def grade_test_output(
+    test_suite: str, instance: pd.Series, test_output: str, test_spec: TestSpec, runtime
+):
+    """
+    Two-pass test grading with short-circuiting:
+    1. Run all tests to identify passing/failing tests
+    2. If no failing tests, evaluate coverage immediately
+    3. Otherwise, run only passing tests for coverage analysis
+    """
+    unit_test_output, coverage_output = '', ''
+    if TESTS_SUFFIX in test_output:
+        unit_test_output = test_output.split(TESTS_SUFFIX)[0]
+
+    if not unit_test_output:
+        return (
+            False,
+            0,
+            '',
+            '',
+            {
+                'total_tests': 0,
+                'passing_tests': 0,
+                'failing_tests': 0,
+                'any_pass': False,
+                'all_pass': False,
+                'passing_test_names': [],
+                'failing_test_names': [],
+            },
+        )
+
+    logger.info('Calling filter unit tests')
+    filtered_content, passing_tests, failing_tests = filter_tests(
+        test_suite, unit_test_output, test_spec.repo
+    )
+
+    total_tests = len(passing_tests) + len(failing_tests)
+    test_stats = {
+        'total_tests': total_tests,
+        'passing_tests': len(passing_tests),
+        'failing_tests': len(failing_tests),
+        'any_pass': len(passing_tests) > 0,
+        'all_pass': len(failing_tests) == 0 and total_tests > 0,
+        'passing_test_names': passing_tests,
+        'failing_test_names': failing_tests,
+    }
+
+    if not passing_tests:
+        return False, 0, unit_test_output, coverage_output, test_stats
+
+    # If all tests pass, evaluate coverage immediately
+    if not failing_tests:
+        coverage = 0
+        cov_success = False
+        if COVERAGE_PREFIX in test_output:
+            coverage_output = test_output.split(COVERAGE_PREFIX)[1]
+            _, coverage = check_coverage(coverage_output, test_spec.code_file)
+            cov_success = True
+        # test_stats['filtered_suite'] = test_suite
+        return cov_success, coverage, unit_test_output, coverage_output, test_stats
+
+    cov_success = False
+    coverage = 0
+    # Second pass - run coverage on passing tests
+    if filtered_content:
+        with tempfile.TemporaryDirectory() as temp_dir:
+            test_suite_path = os.path.join(temp_dir, 'test_suite.py')
+            with open(test_suite_path, 'w') as f:
+                f.write(filtered_content)
+            runtime.copy_to(test_suite_path, '/tmp')
+
+        run_command(runtime, f'cp /tmp/test_suite.py /testbed/{test_spec.test_file}')
+        _, test_output_second_pass, _ = run_tests(runtime, instance, '/tmp/test.sh')
+
+        coverage, coverage_output, unit_test_output = 0, '', test_output_second_pass
+
+        if COVERAGE_PREFIX in test_output_second_pass:
+            coverage_output = test_output_second_pass.split(COVERAGE_PREFIX)[1]
+            unit_test_output = test_output_second_pass.split(TESTS_SUFFIX)[0]
+            _, coverage = check_coverage(coverage_output, test_spec.code_file)
+            cov_success = True
+
+    # test_stats['filtered_suite'] = filtered_content
+    return cov_success, coverage, unit_test_output, coverage_output, test_stats
+
+
+def process_instance(
+    instance: pd.Series,
+    metadata: EvalMetadata,
+    reset_logger: bool = True,
+    log_dir: str | None = None,
+) -> EvalOutput:
+    """
+    Evaluate agent performance on a TestGenEval problem instance.
+
+    Note that this signature differs from the expected input to `run_evaluation`. Use
+    `functools.partial` to provide optional arguments before passing to the evaluation harness.
+
+    Args:
+        log_dir (str | None, default=None): Path to directory where log files will be written. Must
+        be provided if `reset_logger` is set.
+
+    Raises:
+        AssertionError: if the `reset_logger` flag is set without a provided log directory.
+    """
+    if reset_logger:
+        assert (
+            log_dir is not None
+        ), "Can't reset logger without a provided log directory."
+        os.makedirs(log_dir, exist_ok=True)
+        reset_logger_for_multiprocessing(logger, instance.instance_id, log_dir)
+    else:
+        logger.info(f'Starting evaluation for instance {instance.instance_id}.')
+
+    config = get_config(instance)
+    id = instance.instance_id
+    logger.info(f'Starting evaluation for instance {id}.')
+
+    instance['test_result']['id'] = id
+    instance['test_result']['report'] = {
+        'test_output': '',
+        # 'coverage_output': '',
+        # 'mutation_output': '',
+        'empty_generation': False,
+        'error_eval': False,
+        'all_tests_pass': False,
+        'tests_pass': False,
+        'test_timeout': False,
+        'mutation_timeout': False,
+        'coverage_success': False,
+        'mutation_success': False,
+        'coverage': 0,
+        'mutation_score': 0,
+        'mutation_error_interval': -1,
+        'num_mutants': -1,
+    }
+
+    instance['test_result']['lexical'] = {
+        'pred_loc': -1,
+        'gold_loc': -1,
+        'pred_readability': -1,
+        'gold_readability': -1,
+        'pred_methods': -1,
+        'gold_methods': -1,
+        'code_bleu': -1,
+        'bleu': -1,
+        'xmatch': -1,
+        'edit_sim': -1,
+        'rouge_f': -1,
+        'rouge_p': -1,
+        'rouge_r': -1,
+    }
+
+    if instance['test_suite'] == '' or instance['test_suite'] is None:
+        instance['test_result']['report']['empty_generation'] = True
+        return EvalOutput(
+            instance_id=instance.instance_id, test_result=instance['test_result']
+        )
+
+    if not args.skip_lexical:
+        lexical_metrics = compute_lexical_metrics(
+            instance['test_suite'], instance['instance']['test_src']
+        )
+        instance['test_result']['lexical'] = lexical_metrics
+
+    test_suite = instance['test_suite']
+    test_spec: TestSpec = instance['test_spec']
+    runtime = create_runtime(config)
+    call_async_from_sync(runtime.connect)
+    with tempfile.TemporaryDirectory() as temp_dir:
+        test_suite_path = os.path.join(temp_dir, 'test_suite.py')
+        with open(test_suite_path, 'w') as f:
+            f.write(test_suite)
+        runtime.copy_to(test_suite_path, '/tmp')
+
+        test_script_path = os.path.join(temp_dir, 'test.sh')
+        with open(test_script_path, 'w') as f:
+            f.write(test_spec.test_script)
+        runtime.copy_to(test_script_path, '/tmp')
+
+        mutation_script_path = os.path.join(temp_dir, 'mutation.sh')
+        with open(mutation_script_path, 'w') as f:
+            f.write(test_spec.mutation_script)
+        runtime.copy_to(mutation_script_path, '/tmp')
+
+    try:
+        run_command(runtime, 'chmod +x /tmp/test.sh /tmp/mutation.sh')
+        run_command(runtime, f'cp /tmp/test_suite.py /testbed/{test_spec.test_file}')
+
+        # First pass - run all tests
+        _, test_output, test_time = run_tests(runtime, instance, '/tmp/test.sh')
+
+        # Grade tests with two-pass approach
+        coverage_success, coverage, unit_test_output, coverage_output, test_stats = (
+            grade_test_output(test_suite, instance, test_output, test_spec, runtime)
+        )
+
+        # Update report with test statistics
+        instance['test_result']['report'].update(
+            {
+                'test_output': unit_test_output,
+                # 'coverage_output': coverage_output,
+                'tests_pass': test_stats['any_pass'],  # Changed to use any_pass
+                'all_tests_pass': test_stats['all_pass'],  # Added all_pass metric
+                'coverage_success': coverage_success,
+                'coverage': coverage if coverage_success else 0,
+                'test_stats': test_stats,
+            }
+        )
+
+        # Only run mutation testing if we have passing tests and coverage
+        if (
+            not args.skip_mutation
+            and coverage_success
+            and test_stats['any_pass']
+            and coverage > 0
+        ):
+            mutation_timeout = max(10, 1.5 * test_time)
+            mutation_toml = MUTATION_TEMPLATE.format(
+                test_cmd=test_spec.test_cmd,
+                source_fp=test_spec.code_file,
+                timeout=mutation_timeout,
+            )
+
+            with tempfile.TemporaryDirectory() as temp_dir:
+                mutation_toml_path = os.path.join(temp_dir, 'mutation.toml')
+                with open(mutation_toml_path, 'w') as f:
+                    f.write(mutation_toml)
+                runtime.copy_to(mutation_toml_path, '/tmp')
+
+            run_command(runtime, 'cp /tmp/mutation.toml /testbed/mutation.toml')
+
+            mutation_code, mutation_output = run_mutation_testing(
+                runtime, instance, '/tmp/mutation.sh'
+            )
+            # instance['test_result']['report']['mutation_output'] = mutation_output
+            if mutation_output and mutation_code == 0:
+                (
+                    mutation_success,
+                    num_mutants,
+                    mutation_score,
+                    mutation_confidence_interval,
+                ) = check_mutation(mutation_output)
+                instance['test_result']['report']['num_mutants'] = num_mutants
+                instance['test_result']['report']['mutation_success'] = mutation_success
+                instance['test_result']['report']['mutation_score'] = mutation_score
+                instance['test_result']['report']['mutation_error_interval'] = (
+                    mutation_confidence_interval
+                )
+
+        return EvalOutput(
+            instance_id=instance.instance_id, test_result=instance['test_result']
+        )
+    except Exception as e:
+        logger.error(f'Error processing instance {instance.instance_id}: {e}')
+        raise RuntimeError(
+            instance.instance_id,
+            'Unexpected output...',
+            logger,
+        )
+
+    finally:
+        runtime.close()
+
+
+def count_and_log_fields(evaluated_predictions, fields, key):
+    """
+    Count and log the sum of specified fields in the evaluated predictions,
+    ignoring fields with a value of -1. If all values for a field are -1,
+    return -1.
+
+    :param evaluated_predictions: DataFrame containing evaluation results
+    :param fields: List of field names to count
+    :param key: Key to access the field values ('report' or 'lexical')
+    """
+
+    def count_field(row, field):
+        value = row['test_result'][key][field]
+        return (
+            value if value != -1 else None
+        )  # Ignore -1 fields by treating them as None
+
+    for field in fields:
+        # Extract the valid values for the field, ignoring -1
+        valid_values = evaluated_predictions.apply(
+            count_field, args=(field,), axis=1
+        ).dropna()
+
+        if valid_values.empty:  # If all values are -1
+            logger.info(f'# {field}: -1 (All values are -1)')
+        else:
+            count = valid_values.sum()  # Sum of valid values
+            length = len(valid_values)  # Count of valid entries
+            logger.info(f'# {field}: {length}. ({count / length:.2f})')
+
+
+if __name__ == '__main__':
+    parser = get_parser()
+    parser.add_argument(
+        '--input-file', type=str, required=True, help='Path to input predictions file'
+    )
+    parser.add_argument(
+        '--dataset',
+        type=str,
+        default='kjain14/testgeneval',
+        help='Dataset to evaluate on',
+    )
+    parser.add_argument(
+        '--split', type=str, default='test', help='Split to evaluate on'
+    )
+    parser.add_argument(
+        '--skip_mutation', action='store_true', help='Skip mutation testing'
+    )
+    parser.add_argument(
+        '--skip_lexical', action='store_true', help='Skip lexical metrics'
+    )
+    parser.add_argument(
+        '--mutation_timeout',
+        type=int,
+        default=MUTATION_TIMEOUT,
+        help='Mutation timeout',
+    )
+    parser.add_argument(
+        '--mutation_buffer',
+        type=int,
+        default=MUTATION_BUFFER,
+        help='Mutation buffer',
+    )
+    args, _ = parser.parse_known_args()
+
+    dataset: list[TestGenEvalInstance] = load_testgeneval_dataset(
+        args.dataset, args.split
+    )
+
+    logger.info(
+        f'Loaded dataset {args.dataset} with split {args.split} to run inference on.'
+    )
+
+    # Load predictions
+    assert args.input_file.endswith('.jsonl'), 'Input file must be a jsonl file.'
+    predictions = pd.read_json(args.input_file, lines=True)
+    assert (
+        'instance_id' in predictions.columns
+    ), 'Input file must contain instance_id column.'
+
+    if 'test_suite' not in predictions.columns and (
+        'test_result' in predictions.columns
+        and 'test_suite' in predictions['test_result'].iloc(0)
+    ):
+        raise ValueError(
+            'Input file must contain test_suite column OR test_result column with test_suite field.'
+        )
+
+    if 'instance_id_swebench' not in predictions.columns:
+        predictions['instance_id_swebench'] = predictions['instance'].apply(
+            lambda x: x['instance_id_swebench']
+        )
+
+    if 'instance_id' not in predictions.columns and (
+        'instance_id' in predictions['instance'].iloc(0)
+    ):
+        raise ValueError(
+            'Input file must contain id column OR instance column with id field.'
+        )
+
+    if 'instance_id' not in predictions.columns:
+        predictions['instance_id'] = predictions['instance'].apply(
+            lambda x: x['instance_id']
+        )
+
+    if 'test_suite' not in predictions.columns:
+        predictions['test_suite'] = predictions['test_result'].apply(
+            lambda x: x['test_suite']
+        )
+
+    assert len(predictions['instance_id'].unique()) == len(
+        predictions
+    ), 'instance_id column must be unique.'
+
+    assert {'instance_id_swebench', 'test_suite', 'instance_id'}.issubset(
+        set(predictions.columns)
+    ), 'Input file must contain id, instance_id and test_suite columns.'
+
+    predictions['test_spec'] = predictions['instance'].apply(
+        lambda x: make_test_spec(x, args.mutation_timeout, args.mutation_buffer)
+    )
+
+    output_file = args.input_file.replace('.jsonl', '.testgeneval.jsonl')
+    instances = prepare_dataset(predictions, output_file, args.eval_n_limit)
+
+    # If possible, load the relevant metadata to avoid issues with `run_evaluation`.
+    metadata: EvalMetadata | None = None
+    metadata_filepath = os.path.join(os.path.dirname(args.input_file), 'metadata.json')
+    if os.path.exists(metadata_filepath):
+        with open(metadata_filepath, 'r') as metadata_file:
+            data = metadata_file.read()
+            metadata = EvalMetadata.model_validate_json(data)
+
+    # The evaluation harness constrains the signature of `process_instance_func` but we need to
+    # pass extra information. Build a new function object to avoid issues with multiprocessing.
+    process_instance_func = partial(
+        process_instance, log_dir=output_file.replace('.jsonl', '.logs')
+    )
+
+    run_evaluation(
+        instances,
+        metadata=None,
+        output_file=output_file,
+        num_workers=args.eval_num_workers,
+        process_instance_func=process_instance_func,
+    )
+
+    # Load evaluated predictions & print number of resolved predictions
+    evaluated_predictions = pd.read_json(output_file, lines=True)
+    report_fields = [
+        'coverage',
+        'mutation_score',
+        'tests_pass',
+        'all_tests_pass',
+        'empty_generation',
+        'coverage_success',
+        'test_timeout',
+        'error_eval',
+    ]
+    lexical_fields = [
+        'pred_loc',
+        'gold_loc',
+        'pred_methods',
+        'gold_methods',
+        'code_bleu',
+        'bleu',
+        'xmatch',
+        'edit_sim',
+        'rouge_f',
+        'rouge_p',
+        'rouge_r',
+    ]
+
+    # Log report and lexical fields
+    count_and_log_fields(evaluated_predictions, report_fields, key='report')
+    count_and_log_fields(evaluated_predictions, lexical_fields, key='lexical')
@@ -0,0 +1,291 @@
+import re
+
+from evaluation.benchmarks.testgeneval.constants import TestStatus
+
+
+def parse_log_pytest(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with PyTest framework
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    for line in log.split('\n'):
+        if any([line.startswith(x.value) for x in TestStatus]):
+            # Additional parsing for FAILED status
+            if line.startswith(TestStatus.FAILED.value):
+                line = line.replace(' - ', ' ')
+            test_case = line.split()
+            if len(test_case) <= 1:
+                continue
+            test_status_map[test_case[1]] = test_case[0]
+    return test_status_map
+
+
+def parse_log_pytest_options(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with PyTest framework with options
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    option_pattern = re.compile(r'(.*?)\[(.*)\]')
+    test_status_map = {}
+    for line in log.split('\n'):
+        if any([line.startswith(x.value) for x in TestStatus]):
+            # Additional parsing for FAILED status
+            if line.startswith(TestStatus.FAILED.value):
+                line = line.replace(' - ', ' ')
+            test_case = line.split()
+            if len(test_case) <= 1:
+                continue
+            has_option = option_pattern.search(test_case[1])
+            if has_option:
+                main, option = has_option.groups()
+                if (
+                    option.startswith('/')
+                    and not option.startswith('//')
+                    and '*' not in option
+                ):
+                    option = '/' + option.split('/')[-1]
+                test_name = f'{main}[{option}]'
+            else:
+                test_name = test_case[1]
+            test_status_map[test_name] = test_case[0]
+    return test_status_map
+
+
+def parse_log_django(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with Django tester framework
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    lines = log.split('\n')
+
+    prev_test = None
+    for line in lines:
+        line = line.strip()
+
+        # This isn't ideal but the test output spans multiple lines
+        if '--version is equivalent to version' in line:
+            test_status_map['--version is equivalent to version'] = (
+                TestStatus.PASSED.value
+            )
+
+        # Log it in case of error
+        if ' ... ' in line:
+            prev_test = line.split(' ... ')[0]
+
+        pass_suffixes = (' ... ok', ' ... OK', ' ...  OK')
+        for suffix in pass_suffixes:
+            if line.endswith(suffix):
+                # TODO: Temporary, exclusive fix for django__django-7188
+                # The proper fix should involve somehow getting the test results to
+                # print on a separate line, rather than the same line
+                if line.strip().startswith(
+                    'Applying sites.0002_alter_domain_unique...test_no_migrations'
+                ):
+                    line = line.split('...', 1)[-1].strip()
+                test = line.rsplit(suffix, 1)[0]
+                test_status_map[test] = TestStatus.PASSED.value
+                break
+        if ' ... skipped' in line:
+            test = line.split(' ... skipped')[0]
+            test_status_map[test] = TestStatus.SKIPPED.value
+        if line.endswith(' ... FAIL'):
+            test = line.split(' ... FAIL')[0]
+            test_status_map[test] = TestStatus.FAILED.value
+        if line.startswith('FAIL:'):
+            test = line.split()[1].strip()
+            test_status_map[test] = TestStatus.FAILED.value
+        if line.endswith(' ... ERROR'):
+            test = line.split(' ... ERROR')[0]
+            test_status_map[test] = TestStatus.ERROR.value
+        if line.startswith('ERROR:'):
+            test = line.split()[1].strip()
+            test_status_map[test] = TestStatus.ERROR.value
+
+        if line.lstrip().startswith('ok') and prev_test is not None:
+            # It means the test passed, but there's some additional output (including new lines)
+            # between "..." and "ok" message
+            test = prev_test
+            test_status_map[test] = TestStatus.PASSED.value
+
+    # TODO: This is very brittle, we should do better
+    # There's a bug in the django logger, such that sometimes a test output near the end gets
+    # interrupted by a particular long multiline print statement.
+    # We have observed this in one of 3 forms:
+    # - "{test_name} ... Testing against Django installed in {*} silenced.\nok"
+    # - "{test_name} ... Internal Server Error: \/(.*)\/\nok"
+    # - "{test_name} ... System check identified no issues (0 silenced).\nok"
+    patterns = [
+        r'^(.*?)\s\.\.\.\sTesting\ against\ Django\ installed\ in\ ((?s:.*?))\ silenced\)\.\nok$',
+        r'^(.*?)\s\.\.\.\sInternal\ Server\ Error:\ \/(.*)\/\nok$',
+        r'^(.*?)\s\.\.\.\sSystem check identified no issues \(0 silenced\)\nok$',
+    ]
+    for pattern in patterns:
+        for match in re.finditer(pattern, log, re.MULTILINE):
+            test_name = match.group(1)
+            test_status_map[test_name] = TestStatus.PASSED.value
+    return test_status_map
+
+
+def parse_log_pytest_v2(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with PyTest framework (Later Version)
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    escapes = ''.join([chr(char) for char in range(1, 32)])
+    for line in log.split('\n'):
+        line = re.sub(r'\[(\d+)m', '', line)
+        translator = str.maketrans('', '', escapes)
+        line = line.translate(translator)
+        if any([line.startswith(x.value) for x in TestStatus]):
+            if line.startswith(TestStatus.FAILED.value):
+                line = line.replace(' - ', ' ')
+            test_case = line.split()
+            if len(test_case) >= 2:
+                test_status_map[test_case[1]] = test_case[0]
+        # Support older pytest versions by checking if the line ends with the test status
+        elif any([line.endswith(x.value) for x in TestStatus]):
+            test_case = line.split()
+            if len(test_case) >= 2:
+                test_status_map[test_case[0]] = test_case[1]
+    return test_status_map
+
+
+def parse_log_seaborn(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with seaborn testing framework
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    for line in log.split('\n'):
+        if line.startswith(TestStatus.FAILED.value):
+            test_case = line.split()[1]
+            test_status_map[test_case] = TestStatus.FAILED.value
+        elif f' {TestStatus.PASSED.value} ' in line:
+            parts = line.split()
+            if parts[1] == TestStatus.PASSED.value:
+                test_case = parts[0]
+                test_status_map[test_case] = TestStatus.PASSED.value
+        elif line.startswith(TestStatus.PASSED.value):
+            parts = line.split()
+            test_case = parts[1]
+            test_status_map[test_case] = TestStatus.PASSED.value
+    return test_status_map
+
+
+def parse_log_sympy(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with Sympy framework
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    pattern = r'(_*) (.*)\.py:(.*) (_*)'
+    matches = re.findall(pattern, log)
+    for match in matches:
+        test_case = f'{match[1]}.py:{match[2]}'
+        test_status_map[test_case] = TestStatus.FAILED.value
+    for line in log.split('\n'):
+        line = line.strip()
+        if line.startswith('test_'):
+            if line.endswith('[FAIL]') or line.endswith('[OK]'):
+                line = line[: line.rfind('[')]
+                line = line.strip()
+            if line.endswith(' E'):
+                test = line.split()[0]
+                test_status_map[test] = TestStatus.ERROR.value
+            if line.endswith(' F'):
+                test = line.split()[0]
+                test_status_map[test] = TestStatus.FAILED.value
+            if line.endswith(' ok'):
+                test = line.split()[0]
+                test_status_map[test] = TestStatus.PASSED.value
+    return test_status_map
+
+
+def parse_log_matplotlib(log: str) -> dict[str, str]:
+    """
+    Parser for test logs generated with PyTest framework
+
+    Args:
+        log (str): log content
+    Returns:
+        dict: test case to test status mapping
+    """
+    test_status_map = {}
+    for line in log.split('\n'):
+        line = line.replace('MouseButton.LEFT', '1')
+        line = line.replace('MouseButton.RIGHT', '3')
+        if any([line.startswith(x.value) for x in TestStatus]):
+            # Additional parsing for FAILED status
+            if line.startswith(TestStatus.FAILED.value):
+                line = line.replace(' - ', ' ')
+            test_case = line.split()
+            if len(test_case) <= 1:
+                continue
+            test_status_map[test_case[1]] = test_case[0]
+    return test_status_map
+
+
+parse_log_astroid = parse_log_pytest
+parse_log_flask = parse_log_pytest
+parse_log_marshmallow = parse_log_pytest
+parse_log_pvlib = parse_log_pytest
+parse_log_pyvista = parse_log_pytest
+parse_log_sqlfluff = parse_log_pytest
+parse_log_xarray = parse_log_pytest
+
+parse_log_pydicom = parse_log_pytest_options
+parse_log_requests = parse_log_pytest_options
+parse_log_pylint = parse_log_pytest_options
+
+parse_log_astropy = parse_log_pytest_v2
+parse_log_scikit = parse_log_pytest_v2
+parse_log_sphinx = parse_log_pytest_v2
+
+
+MAP_REPO_TO_PARSER = {
+    'astropy/astropy': parse_log_astropy,
+    'django/django': parse_log_django,
+    'marshmallow-code/marshmallow': parse_log_marshmallow,
+    'matplotlib/matplotlib': parse_log_matplotlib,
+    'mwaskom/seaborn': parse_log_seaborn,
+    'pallets/flask': parse_log_flask,
+    'psf/requests': parse_log_requests,
+    'pvlib/pvlib-python': parse_log_pvlib,
+    'pydata/xarray': parse_log_xarray,
+    'pydicom/pydicom': parse_log_pydicom,
+    'pylint-dev/astroid': parse_log_astroid,
+    'pylint-dev/pylint': parse_log_pylint,
+    'pytest-dev/pytest': parse_log_pytest,
+    'pyvista/pyvista': parse_log_pyvista,
+    'scikit-learn/scikit-learn': parse_log_scikit,
+    'sqlfluff/sqlfluff': parse_log_sqlfluff,
+    'sphinx-doc/sphinx': parse_log_sphinx,
+    'sympy/sympy': parse_log_sympy,
+}
@@ -0,0 +1,311 @@
+import sys
+from typing import Callable, Dict, List, Optional, Sequence, TypeVar, Union
+
+import nltk
+import numpy as np
+from fuzzywuzzy import fuzz
+from rouge import Rouge
+
+
+
+# increase recursion depth to ensure ROUGE can be calculated for long sentences
+if sys.getrecursionlimit() < 10_000:
+    sys.setrecursionlimit(10_000)
+
+def bleu(gold: List[str], pred: List[str]) -> float:
+    """
+    Calculate BLEU score, using smoothing method 2 with auto reweighting, in the range of 0~100.
+
+    :param gold: list of gold tokens
+    :param pred: list of predicted tokens
+    :return: BLEU score
+    """
+    if len(pred) == 0 or len(gold) == 0:
+        return 0.0
+    return 100.0 * nltk.translate.bleu_score.sentence_bleu(
+        [gold],
+        pred,
+        smoothing_function=nltk.translate.bleu_score.SmoothingFunction().method2,
+        auto_reweigh=True,
+    )
+
+
+def batch_bleu(golds: List[List[str]], preds: List[List[str]]) -> List[float]:
+    """
+    Calculate BLEU score for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :return: list of BLEU scores
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    return [bleu(gold, pred) for gold, pred in zip(golds, preds)]
+
+
+def corpus_bleu(golds: List[List[str]], preds: List[List[str]]) -> float:
+    """
+    Calculate corpus-level BLEU score for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :return: corpus-level BLEU score
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    return 100.0 * nltk.translate.bleu_score.corpus_bleu(
+        [[gold] for gold in golds],
+        preds,
+        smoothing_function=nltk.translate.bleu_score.SmoothingFunction().method2,
+        auto_reweigh=True,
+    )
+
+
+def edit_sim(
+    gold: Union[str, List[str]], pred: Union[str, List[str]], sep: str = " "
+) -> float:
+    """
+    Calculate char-level edit similarity, in the range of 0~100.
+
+    :param gold: gold sentence or list of gold tokens
+    :param pred: predicted sentence or list of predicted tokens
+    :param sep: separator between tokens
+    :return: char-level edit similarity
+    """
+    if len(pred) == 0 or len(gold) == 0:
+        return 0.0
+    if isinstance(gold, list):
+        gold = sep.join(gold)
+    if isinstance(pred, list):
+        pred = sep.join(pred)
+    return fuzz.ratio(gold, pred)
+
+
+def batch_edit_sim(
+    golds: List[Union[str, List[str]]],
+    preds: List[Union[str, List[str]]],
+    sep: str = " ",
+) -> List[float]:
+    """
+    Calculate char-level edit similarity for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :param sep: separator between tokens
+    :return: list of char-level edit similarity
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    return [edit_sim(gold, pred, sep) for gold, pred in zip(golds, preds)]
+
+
+T = TypeVar("T")
+
+
+def exact_match(gold: T, pred: T) -> float:
+    """
+    Calculate exact match accuracy, in the range of {0, 100}.
+
+    :param gold: gold sentence or list of gold tokens
+    :param pred: predicted sentence or list of predicted tokens
+    :return: exact match accuracy
+    """
+    if len(pred) == 0 or len(gold) == 0:
+        return 0.0
+    return 100.0 if gold == pred else 0.0
+
+
+def batch_exact_match(golds: List[T], preds: List[T]) -> List[float]:
+    """
+    Calculate exact match accuracy for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :return: list of exact match accuracy
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    return [exact_match(gold, pred) for gold, pred in zip(golds, preds)]
+
+
+def rouge_l(
+    gold: Union[str, List[str]], pred: Union[str, List[str]], sep: str = " "
+) -> Dict[str, float]:
+    """
+    Calculate ROUGE-L F1, precision, and recall scores, in the range of 0~100.
+
+    :param gold: gold sentence or list of gold tokens
+    :param pred: predicted sentence or list of predicted tokens
+    :return: {"p": precision, "r": recall, "f": F1}
+    """
+    if len(pred) == 0 or len(gold) == 0:
+        return {"p": 0.0, "r": 0.0, "f": 0.0}
+    if isinstance(gold, list):
+        gold = sep.join(gold)
+    if isinstance(pred, list):
+        pred = sep.join(pred)
+    try:
+        rouge = Rouge()
+        scores = rouge.get_scores(hyps=pred, refs=gold, avg=True)
+        return {x: scores["rouge-l"][x] * 100.0 for x in ["p", "r", "f"]}
+    except ValueError:
+        return {"p": 0.0, "r": 0.0, "f": 0.0}
+
+
+def batch_rouge_l(
+    golds: List[Union[str, List[str]]],
+    preds: List[Union[str, List[str]]],
+    sep: str = " ",
+) -> Dict[str, List[float]]:
+    """
+    Calculate ROUGE-L F1, precision, and recall scores for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :param sep: separator between tokens
+    :return: list of {"p": precision, "r": recall, "f": F1}
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    scores = [rouge_l(gold, pred, sep) for gold, pred in zip(golds, preds)]
+    return {x: [score[x] for score in scores] for x in ["p", "r", "f"]}
+
+
+def accuracy(
+    gold: List[str],
+    pred: List[str],
+    ignore: Optional[Sequence[str]] = None,
+) -> float:
+    """
+    Calculate token-level accuracy, in the range of 0~100.
+    If gold and pred are not the same length, the longer one would be truncated.
+
+    :param gold: list of gold tokens
+    :param pred: list of predicted tokens
+    :param ignore: list of (gold) tokens to ignore
+    :return: accuracy
+    """
+    if len(pred) == 0 or len(gold) == 0:
+        return 0.0
+    if ignore is None:
+        ignore = []
+    i = 0
+    total = 0
+    match = 0
+    while i < len(gold) and i < len(pred):
+        if gold[i] in ignore:
+            i += 1
+            continue
+        total += 1
+        if gold[i] == pred[i]:
+            match += 1
+        i += 1
+
+    if total == 0:
+        return 0.0
+    return 100.0 * match / total
+
+
+def batch_accuracy(
+    golds: List[List[str]],
+    preds: List[List[str]],
+    ignore: Optional[Sequence[str]] = None,
+) -> List[float]:
+    """
+    Calculate token-level accuracy for a batch of sentences.
+
+    :param golds: list of gold sentences
+    :param preds: list of predicted sentences
+    :param ignore: list of (gold) tokens to ignore
+    :return: list of accuracy
+    """
+    if len(golds) != len(preds):
+        raise ValueError("golds and preds must have the same length")
+    return [accuracy(gold, pred, ignore) for gold, pred in zip(golds, preds)]
+
+
+def first_match_to_topk(
+    first_match_list: List[int], k_values: List[int]
+) -> Dict[int, List[float]]:
+    """
+    Calculate top-k accuracy with the first match ranks (1-indexed).
+
+    :param first_match: first match ranks (1-indexed)
+    :param k_values: k values to consider
+    :return: a mapping from k to top-k accuracies (ranging from 0~100)
+    """
+    return {k: [100.0 if x <= k else 0.0 for x in first_match_list] for k in k_values}
+
+
+def pass_at_k(n: int, c: int, k: int) -> float:
+    """
+    Sample pass@k metric according to the Codex paper, but in the scale of 0~100.
+    :param n: total number of samples
+    :param c: number of correct samples
+    :param k: k in pass@$k$
+    """
+    if n < k or (n - c) < k:
+        # fallback to the (1 - (1-p)^k) formula
+        return (1 - (1 - (c / n)) ** k) * 100
+    else:
+        return (1.0 - np.prod(1.0 - k / np.arange(n - c + 1, n + 1)).item()) * 100
+
+
+def self_bleu(samples: List[List[str]]) -> float:
+    """
+    Calculate self-BLEU among the samples.
+    :param samples: the chosen m samples
+    :return: self-BLEU
+    """
+    if len(samples) == 0:
+        return 100.0
+
+    scores = []
+    for i in range(len(samples)):
+        scores.append(
+            100.0
+            * nltk.translate.bleu_score.sentence_bleu(
+                [samples[j] for j in range(len(samples)) if j != i],
+                samples[i],
+                smoothing_function=nltk.translate.bleu_score.SmoothingFunction().method2,
+                auto_reweigh=True,
+            )
+        )
+    return np.mean(scores).item()
+
+
+def self_edit_distance(samples: List[Union[str, List[str]]], sep=" ") -> float:
+    """
+    Calculate self-edit-distance among the samples.
+    :param samples: the chosen m samples
+    :param sep: the separator between tokens
+    :return: self-edit-distance
+    """
+    if len(samples) == 0:
+        return 0.0
+
+    scores = []
+    for i in range(len(samples)):
+        sample_i = samples[i]
+        if not isinstance(sample_i, str):
+            sample_i = sep.join(sample_i)
+        for j in range(len(samples)):
+            if i == j:
+                continue
+            sample_j = samples[j]
+            if not isinstance(sample_j, str):
+                sample_j = sep.join(sample_j)
+
+            scores.append(100 - fuzz.ratio(sample_i, sample_j))
+    return np.mean(scores).item()
+
+
+
+QUALITY_METRICS: Dict[str, Callable[[List[str], List[str]], float]] = {
+    "bleu": bleu,
+    "xmatch": exact_match,
+    "edit-sim": edit_sim,
+    "rouge-f": lambda g, p: rouge_l(g, p)["f"],
+    "rouge-p": lambda g, p: rouge_l(g, p)["p"],
+    "rouge-r": lambda g, p: rouge_l(g, p)["r"],
+}
@@ -0,0 +1,114 @@
+CODEACT_TESTGEN_PROMPT_OLD = """Your goal is to generate a high-quality test suite (at least 20+ passing tests) for the code file: {code_file}. Output the test suite at {test_file}\n'
+
+[current directory: /workspace/{workspace_dir_name}]
+
+IMPORTANT: You should ONLY interact with the environment provided to you AND NEVER ASK FOR HUMAN HELP
+
+IMPORTANT: Follow instructions, if you have < 80 tests you should generate more tests rather than trying to fix the ones you have.
+
+IMPORTANT: Code file to test:
+```python
+{code_src}
+```
+
+Here are additional imports that you may need:
+{imports}
+
+Look at code dependencies (NOT {code_file} since you already have contents) and test files you need context for to write a complete test suite.
+
+Aim for 20+ test functions with asserts. Do not hestitate to use the Python interpreter to understand the input output behavior of the code you are testing.
+
+Output your test suite at {test_file}. Each unit test must be a function starting with test_. Include all your test imports and setup before your first test. Do not include a main method to run the tests. Make sure to make it as comprehensive as possible, try to execute all the methods you saw.
+
+When you think you've successfully generated a test suite, run it on for the current project using {coverage_command}.
+
+If you have few tests GENERATE MORE TESTS rather than trying to fix the ones you have (it is possible to filter out failing tests later).
+
+Then run coverage report -m --include {code_file} to see how well your test suite covers the code under test.
+
+When you are trying to improve coverage pick a part of the code that is not covered (indicated by lines on coverage report), examine the code and then
+try to generate a test for it. Feel free to use a code interpreter to understand the input output behavior. ONLY add tests
+not remove them.
+
+If you are unable to see passing and failing tests, FIX YOUR IMPORTS to use the same style as other test files.
+
+You should NOT modify any existing test case files. You SHOULD add new test in a NEW file to reproduce the issue.
+
+You should NEVER use web browsing or any other web-based tools.
+
+You should NEVER install new packages, use existing packages only.
+
+You should ALWAYS use the default Python interpreter available in the <execute_bash> environment to run code related to the provided issue and/or repository.
+
+You should ALWAYS use local imports DO NOT import the general library.
+
+When you think you have a fully adequate test suite, please run the following command: <execute_bash> exit </execute_bash>.
+"""
+
+CODEACT_TESTGEN_PROMPT = """
+Your goal is to generate a comprehensive, **broad-coverage** test suite for the code below, ensuring you test as many lines and branches as possible on the first attempt.
+
+Place your test suite in a new file named {test_file}.
+
+IMPORTANT REQUIREMENTS:
+1. **No external help or resources**—use only the snippet below.
+2. **Focus on breadth over depth**: cover all major functions, classes, and code paths early to minimize coverage iterations.
+3. Each test function must start with `test_` and use `assert` to verify behavior.
+4. Include only necessary imports (standard library or local).
+5. Do **not** modify existing test files—create a brand new one. No `main()` or other non-test code.
+6. Produce **at least 20 test functions**; if coverage is lacking, add more tests rather than removing or changing existing ones.
+7. Use the following commands to check coverage:
+   <execute_bash> {coverage_command} </execute_bash>
+   <execute_bash> coverage report -m --include {code_file} </execute_bash>
+   If lines remain uncovered, add new tests targeting them specifically.
+8. When you're satisfied with coverage, finalize by running:
+   <execute_bash> exit </execute_bash>
+
+Below is the **complete code snippet** to test:
+
+<START_OF_CODE>
+{code_src}
+<END_OF_CODE>
+
+NOTE: if you are testing django, you must use from django.test import SimpleTestCase and class based tests (i.e. class TestSomething(SimpleTestCase)).
+NOTE: if there is an error executing tests you MUST fix it before exiting. DO NOT install new packages.
+NOTE: if outputting a revised test suite REPLACE {test_file} with the revised suite
+
+**Output the final test suite** (20+ tests) for {test_file} in a single code block, no extra commentary. MAKE SURE you run the tests and ensure you can see which tests passed and failed BEFORE exiting.
+"""
+
+CODEACT_TESTGEN_PROMPT_ITERATE = """
+Your goal is to improve the test suite at {test_file} to achieve **broad-coverage** of the code below.
+
+First run the test suite.
+
+If no tests run, then remove {test_file} and create {test_file} with a new suite.
+
+Otherwise, improve it aiming to improve code coverage.
+
+IMPORTANT REQUIREMENTS:
+1. Use the following commands to check coverage (RUN THIS FIRST):
+   <execute_bash> {coverage_command} </execute_bash>
+   <execute_bash> coverage report -m --include {code_file} </execute_bash>
+   If lines remain uncovered, add new tests targeting them specifically.
+2. **No external help or resources**—use only the snippet below.
+3. **Focus on breadth over depth**: cover all major functions, classes, and code paths early to minimize coverage iterations.
+4. Each test function must use `assert` to verify behavior.
+5. Include only necessary imports (standard library or local).
+6. Do **not** modify other test files in the repository. No `main()` or other non-test code.
+7. Produce **at least 20 test functions**; if coverage is lacking, add more tests rather than removing or changing existing ones.
+8. When you're satisfied with coverage, finalize by running:
+   <execute_bash> exit </execute_bash>
+
+Below is the **complete code snippet** to test:
+
+<START_OF_CODE>
+{code_src}
+<END_OF_CODE>
+
+NOTE: if you are testing django, you must use from django.test import SimpleTestCase and class based tests (i.e. class TestSomething(SimpleTestCase)).
+NOTE: if there is an error executing tests you MUST fix it before exiting. DO NOT install new packages.
+NOTE: if outputting a revised test suite REPLACE {test_file} with the revised suite
+
+**Output the final test suite** (20+ tests) for {test_file} in a single code block, no extra commentary. MAKE SURE you run the tests and ensure you can see which tests passed and failed BEFORE exiting.
+"""
@@ -0,0 +1,31 @@
+import re
+from pygments.lexers.python import PythonLexer
+
+def tokenize_code(code):
+    lexer = PythonLexer()
+    tokens = process_pygments_tokens(lexer.get_tokens(code))
+    return tokens
+
+def process_pygments_tokens(tokens):
+    new_tokens = []
+
+    for token in tokens:
+        if str(token[0]) == "Token.Text" and re.match(r'\s+', token[1]) or str(token[0]) == "Token.Text.Whitespace":
+            continue
+        new_tokens.append(token[1])
+
+    new_tokens_final = []
+    i = 0
+    while i < len(new_tokens)-2:
+        if new_tokens[i] == '"' and new_tokens[i+1]=='STR' and new_tokens[i+2] == '"':
+            new_tokens_final.append("\"STR\"")
+            i = i + 3
+        else:
+            new_tokens_final.append(new_tokens[i])
+            i = i + 1
+    
+    for i in range(len(new_tokens)-2, len(new_tokens)):
+        if i >= 0:
+            new_tokens_final.append(new_tokens[i])
+
+    return new_tokens_final
@@ -0,0 +1,58 @@
+import json
+import re
+
+
+def check_coverage(coverage_output, code_file):
+    json_cov = json.loads(coverage_output)
+    if code_file in json_cov['files'].keys():
+        file_data = json_cov['files'][code_file]
+        return True, file_data['summary']['percent_covered']
+
+    return False, 0
+
+
+def check_mutation(mutation_output):
+    if 'total jobs: ' in mutation_output:
+        num_mutants = int(mutation_output.split('total jobs: ')[1].split('\n')[0])
+        final_conf = mutation_output.split('\n')[-1]
+        if len(final_conf.strip().split(' ')) == 3:
+            low, val, high = final_conf.split(' ')
+            low = float(low)
+            val = float(val)
+            high = float(high)
+
+            confidence_range = high - val
+            mutation_score = 100 - val
+
+            return True, num_mutants, mutation_score, confidence_range
+
+    return False, -1, 0, -1
+
+
+def count_methods(code_str):
+    """
+    Counts the number of methods/functions in a given string of code.
+
+    Args:
+    code_str (str): A string containing code.
+
+    Returns:
+    int: The number of methods/functions found.
+    """
+    # Regular expression to find Python function definitions
+    pattern = r'\bdef\b\s+\w+\s*\('
+    matches = re.findall(pattern, code_str)
+    return len(matches)
+
+
+def get_lines_of_code(code_str):
+    """
+    Extracts lines of code from a given string.
+
+    Args:
+    code_str (str): A string containing code.
+
+    Returns:
+    list: A list of lines of code.
+    """
+    return len(code_str.strip().split('\n'))
@@ -0,0 +1,577 @@
+import asyncio
+import json
+import os
+import tempfile
+import time
+import traceback
+from typing import Any
+
+import numpy as np
+import pandas as pd
+import toml
+from datasets import load_dataset
+
+import openhands.agenthub
+from evaluation.benchmarks.testgeneval.constants import MAP_REPO_VERSION_TO_SPECS
+from evaluation.benchmarks.testgeneval.prompt import (
+    CODEACT_TESTGEN_PROMPT,
+    CODEACT_TESTGEN_PROMPT_ITERATE,
+)
+from evaluation.benchmarks.testgeneval.utils import get_test_directives
+from evaluation.utils.shared import (
+    EvalException,
+    EvalMetadata,
+    EvalOutput,
+    assert_and_raise,
+    codeact_user_response,
+    get_metrics,
+    is_fatal_evaluation_error,
+    make_metadata,
+    prepare_dataset,
+    reset_logger_for_multiprocessing,
+    run_evaluation,
+    update_llm_config_for_completions_logging,
+)
+from openhands.controller.state.state import State
+from openhands.core.config import (
+    AgentConfig,
+    AppConfig,
+    SandboxConfig,
+    get_llm_config_arg,
+    get_parser,
+)
+from openhands.core.logger import openhands_logger as logger
+from openhands.core.main import create_runtime, run_controller
+from openhands.events.action import CmdRunAction, MessageAction
+from openhands.events.observation import CmdOutputObservation, ErrorObservation
+from openhands.events.serialization.event import event_to_dict
+from openhands.runtime.base import Runtime
+from openhands.utils.async_utils import call_async_from_sync
+
+RUN_WITH_BROWSING = os.environ.get('RUN_WITH_BROWSING', 'false').lower() == 'true'
+
+AGENT_CLS_TO_FAKE_USER_RESPONSE_FN = {
+    'CodeActAgent': codeact_user_response,
+}
+
+
+def _preprocess_instance(d):
+    for key, value in d.items():
+        if isinstance(value, np.ndarray):
+            d[key] = value.tolist()
+    return d
+
+
+def _get_swebench_workspace_dir_name(instance: pd.Series) -> str:
+    return f'{instance.repo}__{instance.version}'.replace('/', '__')
+
+
+def get_instruction(instance: pd.Series, metadata: EvalMetadata):
+    # workspace_dir_name = _get_swebench_workspace_dir_name(instance)
+    # Prepare instruction
+    coverage_command = ' '.join(
+        [
+            MAP_REPO_VERSION_TO_SPECS[instance['repo']][instance['version']][
+                'test_cmd'
+            ],
+            *get_test_directives(instance),
+        ]
+    )
+
+    # Testing general agents
+    prompt_to_use = (
+        CODEACT_TESTGEN_PROMPT_ITERATE
+        if instance['full_pred'] is not None
+        else CODEACT_TESTGEN_PROMPT
+    )
+    instruction = prompt_to_use.format(
+        code_file=os.path.join('/testbed', instance.code_file),
+        test_file=os.path.join('/testbed', instance.test_file),
+        coverage_command=coverage_command,
+        code_src=instance['code_src'],
+        imports='\n'.join(instance.local_imports),
+        workspace_dir_name=_get_swebench_workspace_dir_name(instance),
+    )
+
+    if RUN_WITH_BROWSING:
+        instruction += (
+            '<IMPORTANT!>\n'
+            'You SHOULD NEVER attempt to browse the web. '
+            '</IMPORTANT!>\n'
+        )
+
+    return instruction
+
+
+# TODO: migrate all swe-bench docker to ghcr.io/openhands
+DOCKER_IMAGE_PREFIX = os.environ.get('EVAL_DOCKER_IMAGE_PREFIX', 'docker.io/kdjain/')
+logger.info(f'Using docker image prefix: {DOCKER_IMAGE_PREFIX}')
+
+
+def get_instance_docker_image(instance_id: str) -> str:
+    image_name = 'sweb.eval.x86_64.' + instance_id
+    image_name = image_name.replace(
+        '__', '_s_'
+    )  # to comply with docker image naming convention
+    return DOCKER_IMAGE_PREFIX.rstrip('/') + '/' + image_name
+
+
+def get_config(
+    instance: pd.Series,
+    metadata: EvalMetadata,
+) -> AppConfig:
+    # We use a different instance image for the each instance of TestGenEval
+    base_container_image = get_instance_docker_image(instance['instance_id_swebench'])
+    logger.info(
+        f'Using instance container image: {base_container_image}. '
+        f'Please make sure this image exists. '
+        f'Submit an issue on https://github.com/All-Hands-AI/OpenHands if you run into any issues.'
+    )
+
+    config = AppConfig(
+        default_agent=metadata.agent_class,
+        run_as_openhands=False,
+        max_iterations=metadata.max_iterations,
+        runtime=os.environ.get('RUNTIME', 'eventstream'),
+        sandbox=SandboxConfig(
+            base_container_image=base_container_image,
+            enable_auto_lint=True,
+            use_host_network=False,
+            # large enough timeout, since some testcases take very long to run
+            timeout=300,
+            # Add platform to the sandbox config to solve issue 4401
+            platform='linux/amd64',
+            api_key=os.environ.get('ALLHANDS_API_KEY', None),
+            remote_runtime_api_url=os.environ.get(
+                'SANDBOX_REMOTE_RUNTIME_API_URL', 'http://localhost:8000'
+            ),
+            keep_runtime_alive=False,
+            remote_runtime_init_timeout=3600,
+        ),
+        # do not mount workspace
+        workspace_base=None,
+        workspace_mount_path=None,
+    )
+    config.set_llm_config(
+        update_llm_config_for_completions_logging(
+            metadata.llm_config, metadata.eval_output_dir, instance['id']
+        )
+    )
+    agent_config = AgentConfig(
+        codeact_enable_jupyter=False,
+        codeact_enable_browsing=RUN_WITH_BROWSING,
+        codeact_enable_llm_editor=False,
+        condenser=metadata.condenser_config,
+    )
+    config.set_agent_config(agent_config)
+    return config
+
+
+def initialize_runtime(
+    runtime: Runtime,
+    instance: pd.Series,  # this argument is not required
+):
+    """Initialize the runtime for the agent.
+
+    This function is called before the runtime is used to run the agent.
+    """
+    logger.info('-' * 30)
+    logger.info('BEGIN Runtime Initialization Fn')
+    logger.info('-' * 30)
+    workspace_dir_name = _get_swebench_workspace_dir_name(instance)
+    obs: CmdOutputObservation
+
+    instance['instance_id'] = instance['instance_id_swebench']
+
+    # Set instance id
+    action = CmdRunAction(
+        command=f"""echo 'export SWE_INSTANCE_ID={instance['instance_id_swebench']}' >> ~/.bashrc && echo 'export PIP_CACHE_DIR=~/.cache/pip' >> ~/.bashrc && echo "alias git='git --no-pager'" >> ~/.bashrc"""
+    )
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(
+        obs.exit_code == 0, f'Failed to export SWE_INSTANCE_ID: {str(obs)}'
+    )
+
+    action = CmdRunAction(command="""export USER=$(whoami); echo USER=${USER} """)
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(obs.exit_code == 0, f'Failed to export USER: {str(obs)}')
+
+    # inject the init script
+    script_dir = os.path.dirname(__file__)
+
+    # inject the instance info
+    action = CmdRunAction(command='mkdir -p /swe_util/eval_data/instances')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(
+        obs.exit_code == 0,
+        f'Failed to create /swe_util/eval_data/instances: {str(obs)}',
+    )
+
+    swe_instance_json_name = 'swe-bench-instance.json'
+    swe_prediction = 'test_suite.py'
+    with tempfile.TemporaryDirectory() as temp_dir:
+        # Construct the full path for the desired file name within the temporary directory
+        temp_file_path = os.path.join(temp_dir, swe_instance_json_name)
+        # Write to the file with the desired name within the temporary directory
+        with open(temp_file_path, 'w') as f:
+            if not isinstance(instance, dict):
+                preprocessed_instance = _preprocess_instance(instance.to_dict())
+                json.dump([preprocessed_instance], f)
+            else:
+                preprocessed_instance = _preprocess_instance(instance)
+                json.dump([preprocessed_instance], f)
+
+        # Copy the file to the desired location
+        runtime.copy_to(temp_file_path, '/swe_util/eval_data/instances/')
+
+        if instance['full_pred'] is not None:
+            temp_file_path_pred = os.path.join(temp_dir, swe_prediction)
+            with open(temp_file_path_pred, 'w') as f:
+                f.write(instance['full_pred'])
+
+            runtime.copy_to(temp_file_path_pred, '/tmp')
+
+            # Copy the file to the desired location
+            action = CmdRunAction(
+                command=f"cp /tmp/test_suite.py /testbed/{instance['test_file']}"
+            )
+            action.set_hard_timeout(600)
+            logger.info(action, extra={'msg_type': 'ACTION'})
+            obs = runtime.run_action(action)
+            logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+            assert_and_raise(
+                obs.exit_code == 0, f'Failed to copy test file: {str(obs)}'
+            )
+
+            action = CmdRunAction(
+                command='git -C /testbed add . && git -C /testbed commit -m "Add test file"'
+            )
+            action.set_hard_timeout(600)
+            logger.info(action, extra={'msg_type': 'ACTION'})
+            obs = runtime.run_action(action)
+            logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+            assert_and_raise(obs.exit_code == 0, f'Failed to cat ~/.bashrc: {str(obs)}')
+
+    # inject the instance swe entry
+    runtime.copy_to(
+        str(os.path.join(script_dir, 'scripts/setup/instance_swe_entry.sh')),
+        '/swe_util/',
+    )
+    action = CmdRunAction(command='cat ~/.bashrc')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(obs.exit_code == 0, f'Failed to cat ~/.bashrc: {str(obs)}')
+
+    action = CmdRunAction(command='source ~/.bashrc')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    if isinstance(obs, ErrorObservation):
+        logger.error(f'Failed to source ~/.bashrc: {str(obs)}')
+    assert_and_raise(obs.exit_code == 0, f'Failed to source ~/.bashrc: {str(obs)}')
+
+    action = CmdRunAction(command='source /swe_util/instance_swe_entry.sh')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(
+        obs.exit_code == 0,
+        f'Failed to source /swe_util/instance_swe_entry.sh: {str(obs)}',
+    )
+
+    action = CmdRunAction(command=f'cd /workspace/{workspace_dir_name}')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(
+        obs.exit_code == 0,
+        f'Failed to cd to /workspace/{workspace_dir_name}: {str(obs)}',
+    )
+
+    action = CmdRunAction(command='git reset --hard')
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(obs.exit_code == 0, f'Failed to git reset --hard: {str(obs)}')
+
+    action = CmdRunAction(
+        command='for remote_name in $(git remote); do git remote remove "${remote_name}"; done'
+    )
+    action.set_hard_timeout(600)
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert_and_raise(obs.exit_code == 0, f'Failed to remove git remotes: {str(obs)}')
+
+    logger.info('-' * 30)
+    logger.info('END Runtime Initialization Fn')
+    logger.info('-' * 30)
+
+
+def complete_runtime(
+    runtime: Runtime,
+    instance: pd.Series,  # this argument is not required, but it is used to get the workspace_dir_name
+) -> dict[str, Any]:
+    """Complete the runtime for the agent.
+
+    This function is called before the runtime is used to run the agent.
+    If you need to do something in the sandbox to get the correctness metric after
+    the agent has run, modify this function.
+    """
+    try:
+        logger.info('-' * 30)
+        logger.info('BEGIN Runtime Completion Fn')
+        logger.info('-' * 30)
+        obs: CmdOutputObservation
+        workspace_dir_name = _get_swebench_workspace_dir_name(instance)
+
+        action = CmdRunAction(command=f'cd /workspace/{workspace_dir_name}')
+        action.set_hard_timeout(600)
+        logger.info(action, extra={'msg_type': 'ACTION'})
+        obs = runtime.run_action(action)
+        logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+        assert_and_raise(
+            obs.exit_code == 0,
+            f'Failed to cd to /workspace/{workspace_dir_name}: {str(obs)}',
+        )
+
+        action = CmdRunAction(command=f'cat {instance.test_file}')
+        action.set_hard_timeout(600)
+        logger.info(action, extra={'msg_type': 'ACTION'})
+        obs = runtime.run_action(action)
+        logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+        assert_and_raise(
+            obs.exit_code == 0,
+            f'Failed to find file: {instance.test_file} in /workspace/{workspace_dir_name}',
+        )
+
+        test_suite = obs.content.strip()
+    except Exception:
+        # Print stack trace
+        print('Skipping, exception in complete_runtime')
+        print(traceback.format_exc())
+        test_suite = instance['full_pred'] if instance['full_pred'] is not None else ''
+
+    # action = CmdRunAction(command='git add -A')
+    # action.set_hard_timeout(600)
+    # logger.info(action, extra={'msg_type': 'ACTION'})
+    # obs = runtime.run_action(action)
+    # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    # assert_and_raise(obs.exit_code == 0, f'Failed to git add -A: {str(obs)}')
+
+    logger.info('-' * 30)
+    logger.info('END Runtime Completion Fn')
+    logger.info('-' * 30)
+    return {
+        'test_suite': test_suite,
+    }
+
+
+def process_instance(
+    instance: pd.Series,
+    metadata: EvalMetadata,
+    reset_logger: bool = True,
+) -> EvalOutput:
+    config = get_config(instance, metadata)
+    start_time = time.time()  # Track start time
+
+    # Setup the logger properly, so you can run multi-processing to parallelize the evaluation
+    if reset_logger:
+        log_dir = os.path.join(metadata.eval_output_dir, 'infer_logs')
+        reset_logger_for_multiprocessing(logger, instance.id, log_dir)
+    else:
+        logger.info(f'Starting evaluation for instance {instance.id}.')
+
+    runtime = create_runtime(config)
+    call_async_from_sync(runtime.connect)
+
+    try:
+        initialize_runtime(runtime, instance)
+
+        instruction = get_instruction(instance, metadata)
+
+        # Here's how you can run the agent (similar to the `main` function) and get the final task state
+        state: State | None = asyncio.run(
+            run_controller(
+                config=config,
+                initial_user_action=MessageAction(content=instruction),
+                runtime=runtime,
+                fake_user_response_fn=AGENT_CLS_TO_FAKE_USER_RESPONSE_FN[
+                    metadata.agent_class
+                ],
+            )
+        )
+
+        # if fatal error, throw EvalError to trigger re-run
+        if is_fatal_evaluation_error(state.last_error):
+            raise EvalException('Fatal error detected: ' + state.last_error)
+
+        # ======= THIS IS SWE-Bench specific =======
+        return_val = complete_runtime(runtime, instance)
+        test_suite = return_val['test_suite']
+        logger.info(
+            f'Got test suite for instance {instance.instance_id}:\n--------\n{test_suite}\n--------'
+        )
+    finally:
+        runtime.close()
+
+    end_time = time.time()
+    elapsed_time = end_time - start_time
+    logger.info(
+        f'Evaluation for instance {instance.instance_id} took {elapsed_time:.2f} seconds.'
+    )
+
+    # ==========================================
+
+    # ======= Attempt to evaluate the agent's edits =======
+    # we use eval_infer.sh to evaluate the agent's edits, not here
+    # because the agent may alter the environment / testcases
+    test_result = {
+        'test_suite': test_suite,
+        'elapsed_time': elapsed_time,
+    }
+
+    # If you are working on some simpler benchmark that only evaluates the final model output (e.g., in a MessageAction)
+    # You can simply get the LAST `MessageAction` from the returned `state.history` and parse it for evaluation.
+    if state is None:
+        raise ValueError('State should not be None.')
+
+    histories = [event_to_dict(event) for event in state.history]
+    metrics = get_metrics(state)
+
+    # Save the output
+    output = EvalOutput(
+        instance_id=instance.id,
+        instruction=instruction,
+        instance=_preprocess_instance(instance.to_dict()),  # SWE Bench specific
+        test_result=test_result,
+        metadata=metadata,
+        history=histories,
+        metrics=metrics,
+        error=state.last_error if state and state.last_error else None,
+    )
+    # print(output)
+    return output
+
+
+def prepare_dataset_pre(dataset: pd.DataFrame, filter_column: str) -> pd.DataFrame:
+    file_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'config.toml')
+    if os.path.exists(file_path):
+        with open(file_path, 'r') as file:
+            data = toml.load(file)
+            if 'selected_ids' in data:
+                selected_ids = data['selected_ids']
+                logger.info(
+                    f'Filtering {len(selected_ids)} tasks from "selected_ids"...'
+                )
+                subset = dataset[dataset[filter_column].isin(selected_ids)]
+                logger.info(f'Retained {subset.shape[0]} tasks after filtering')
+
+                subset['instance_id_swebench'] = subset['instance_id']
+                subset['instance_id'] = subset['id']
+                return subset
+
+    dataset['instance_id_swebench'] = dataset['instance_id']
+    dataset['instance_id'] = dataset['id']
+    return dataset
+
+
+if __name__ == '__main__':
+    parser = get_parser()
+    parser.add_argument(
+        '--dataset',
+        type=str,
+        default='kjain/testgenevallite',
+        help='data set to evaluate on, either full-test or lite-test',
+    )
+    parser.add_argument(
+        '--split',
+        type=str,
+        default='test',
+        help='split to evaluate on',
+    )
+    parser.add_argument(
+        '--testfile_start',
+        action='store_true',
+        help='Whether to start from the 0 shot test file',
+    )
+
+    parser.add_argument(
+        '--zero_shot_path',
+        type=str,
+        help='Path to the zero shot test file predictions',
+    )
+    args, _ = parser.parse_known_args()
+
+    if args.testfile_start and not args.zero_shot_path:
+        raise ValueError(
+            'If you want to start from the 0 shot test file, you must provide the path to the zero shot test file predictions'
+        )
+
+    preds_map = {}
+    if args.testfile_start:
+        with open(args.zero_shot_path, 'r') as f:
+            for line in f:
+                pred = json.loads(line)
+                preds_map[pred['id']] = pred['preds']['full'][0]
+
+    # NOTE: It is preferable to load datasets from huggingface datasets and perform post-processing
+    # so we don't need to manage file uploading to OpenHands's repo
+    dataset = load_dataset(args.dataset, split=args.split)
+    logger.info(f'Loaded dataset {args.dataset} with split {args.split}')
+    testgeneval_filepairs = prepare_dataset_pre(dataset.to_pandas(), 'id')
+
+    llm_config = None
+    if args.llm_config:
+        llm_config = get_llm_config_arg(args.llm_config)
+        llm_config.log_completions = True
+        # modify_params must be False for evaluation purpose, for reproducibility and accurancy of results
+        llm_config.modify_params = False
+
+    if llm_config is None:
+        raise ValueError(f'Could not find LLM config: --llm_config {args.llm_config}')
+
+    details = {}
+    _agent_cls = openhands.agenthub.Agent.get_cls(args.agent_cls)
+
+    dataset_descrption = (
+        args.dataset.replace('/', '__') + '-' + args.split.replace('/', '__')
+    )
+    metadata = make_metadata(
+        llm_config,
+        dataset_descrption,
+        args.agent_cls,
+        args.max_iterations,
+        args.eval_note,
+        args.eval_output_dir,
+        details=details,
+    )
+
+    output_file = os.path.join(metadata.eval_output_dir, 'output.jsonl')
+    instances = prepare_dataset(testgeneval_filepairs, output_file, args.eval_n_limit)
+
+    if not instances.empty:
+        instances['full_pred'] = (
+            instances['instance_id']
+            .map(preds_map)
+            .apply(lambda x: x if pd.notna(x) else None)
+        )
+
+        run_evaluation(
+            instances, metadata, output_file, args.eval_num_workers, process_instance
+        )
@@ -0,0 +1,128 @@
+import argparse
+import os
+import subprocess
+
+from datasets import load_dataset
+
+
+# Function to run shell commands
+def run_command(command):
+    try:
+        subprocess.run(command, check=True, shell=True)
+    except subprocess.CalledProcessError as e:
+        print(f'An error occurred: {e}')
+
+
+# Function to log in to Docker Hub
+def docker_login():
+    print('Logging into Docker Hub...')
+    run_command('docker login')
+
+
+# Function to generate Dockerfile content based on image type
+def generate_dockerfile_content(
+    base_image, dependencies, datum, patch_path, test_patch_path
+):
+    dockerfile_content = f"""
+FROM {base_image}
+SHELL ["/bin/bash", "-c"]
+RUN source /opt/miniconda3/bin/activate && conda activate testbed && pip install {' '.join(dependencies)}
+COPY {patch_path} /app/patch.diff
+RUN git apply /app/patch.diff
+RUN rm /app/patch.diff
+COPY {test_patch_path} /app/patch.diff
+RUN git apply /app/patch.diff
+RUN git config --global user.email ""
+RUN git config --global user.name "TestGenEval"
+RUN rm /app/patch.diff
+RUN rm {datum['test_file']}
+"""
+
+    # Add specific content based on image type
+    dockerfile_content += 'RUN git add .\nRUN git commit -m "Testing fixes"'
+
+    return dockerfile_content
+
+
+# Function to build, push, and clean up Docker images
+def build_and_push_image(dockerfile_content, image_name):
+    with open('Dockerfile.temp', 'w') as dockerfile:
+        dockerfile.write(dockerfile_content)
+    run_command(f'docker build -f Dockerfile.temp -t {image_name} .')
+    run_command(f'docker push {image_name}')
+    run_command(f'docker rmi {image_name}')
+    os.remove('Dockerfile.temp')
+
+
+# Function to process images with .eval in the name
+def process_images(dataset, original_namespace, new_namespace, start_instance_id):
+    dependencies = ['coverage', 'cosmic-ray']
+
+    found_start = len(start_instance_id) == 0
+    for datum in dataset:
+        if not found_start and datum['instance_id'] == start_instance_id:
+            found_start = True
+        elif found_start:
+            full_image_name = f'{original_namespace}/sweb.eval.x86_64.{datum["instance_id"].replace("__", "_s_")}:latest'
+            print(f'Processing image: {full_image_name}')
+            run_command(f'docker pull {full_image_name}')
+
+            # Save patches and preds_context to regular files
+            patch_file_path = 'patch.diff'
+            test_patch_file_path = 'test_patch.diff'
+
+            with open(patch_file_path, 'w') as patch_file, open(
+                test_patch_file_path, 'w'
+            ) as test_patch_file:
+                patch_file.write(datum['patch'])
+                test_patch_file.write(datum['test_patch'])
+
+            # Define image types and corresponding tags
+            new_image_name = f'{new_namespace}/sweb.eval.x86_64.{datum["instance_id"].replace("__", "_s_")}:latest'
+            dockerfile_content = generate_dockerfile_content(
+                full_image_name,
+                dependencies,
+                datum,
+                patch_file_path,
+                test_patch_file_path,
+            )
+            build_and_push_image(dockerfile_content, new_image_name)
+
+            # Cleanup regular files and images
+            os.remove(patch_file_path)
+            os.remove(test_patch_file_path)
+            run_command(f'docker rmi {full_image_name}')
+            run_command('docker system prune -f')  # Clean up dangling resources
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(
+        description='Process Docker images with .eval in the name.'
+    )
+    parser.add_argument('--dataset', type=str, default='kjain14/testgeneval')
+    parser.add_argument('--split', type=str, default='test')
+    parser.add_argument(
+        '--new_namespace',
+        type=str,
+        default='kdjain',
+        help='The new Docker Hub namespace to push the images',
+    )
+    parser.add_argument(
+        '--original_namespace',
+        type=str,
+        default='xingyaoww',
+        help='The original Docker Hub namespace',
+    )
+    parser.add_argument(
+        '--start_instance_id',
+        type=str,
+        default='',
+        help='The instance_id to start processing from',
+    )
+    args = parser.parse_args()
+    dataset = load_dataset(args.dataset)[args.split]
+
+    docker_login()
+    process_images(
+        dataset, args.original_namespace, args.new_namespace, args.start_instance_id
+    )
@@ -0,0 +1,196 @@
+sweb.base.x86_64:latest
+sweb.env.x86_64.088a7e628bda9770f9757b:latest
+sweb.env.x86_64.0d80c7dec81ee2f2f513e2:latest
+sweb.env.x86_64.0f99bce2750f3109957bec:latest
+sweb.env.x86_64.1b3b218535da0abf4469cb:latest
+sweb.env.x86_64.1c1a6945f732f9391228c5:latest
+sweb.env.x86_64.1f92e6d7cef88badc4f744:latest
+sweb.env.x86_64.27dd9791e13f5c857a09f9:latest
+sweb.env.x86_64.297af196949a2a635bce66:latest
+sweb.env.x86_64.2baaea72acc974f6c02079:latest
+sweb.env.x86_64.2e50125951bc69cddd7421:latest
+sweb.env.x86_64.2f217c8b4490bfa0e2ba14:latest
+sweb.env.x86_64.31244378a92e3bcce809ac:latest
+sweb.env.x86_64.428468730904ff6b4232aa:latest
+sweb.env.x86_64.5d1fda9d55d65d8a4e5bdb:latest
+sweb.env.x86_64.6b007979cf533f0f3016e8:latest
+sweb.env.x86_64.7037e8c448a4b8ebfe9b13:latest
+sweb.env.x86_64.71498c7426dbf05599642f:latest
+sweb.env.x86_64.756beac07713d7e8dc1129:latest
+sweb.env.x86_64.78278ae2cf880e395f1337:latest
+sweb.env.x86_64.8f1f7b974f0c57c7aeba39:latest
+sweb.env.x86_64.934a137824256b612e9dc5:latest
+sweb.env.x86_64.a0efca7a0fe6719dbf65c2:latest
+sweb.env.x86_64.a18371b03f944585b4f08c:latest
+sweb.env.x86_64.a33dddf55cdff5d8e23374:latest
+sweb.env.x86_64.aa92880033da20ca313928:latest
+sweb.env.x86_64.b649f0ff62fad147f7f073:latest
+sweb.env.x86_64.b7ce4be3b3c35f68c61248:latest
+sweb.env.x86_64.c70909fdac4897d1c685df:latest
+sweb.env.x86_64.c795f4b88616b8462021ed:latest
+sweb.env.x86_64.cc47cc71483942d0c3a15e:latest
+sweb.env.x86_64.dc5ff4c0e3fe8db5afc4da:latest
+sweb.env.x86_64.e3afd7f04b325a4de4982d:latest
+sweb.env.x86_64.e5bb89bf78258a7d14c34b:latest
+sweb.env.x86_64.e83e37f52c09532c62acfb:latest
+sweb.env.x86_64.efa6065ed5bf204410fd53:latest
+sweb.eval.x86_64.django_s_django-17087:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-10508:latest
+sweb.eval.x86_64.django_s_django-14017:latest
+sweb.eval.x86_64.django_s_django-11422:latest
+sweb.eval.x86_64.sympy_s_sympy-14774:latest
+sweb.eval.x86_64.django_s_django-14915:latest
+sweb.eval.x86_64.sympy_s_sympy-22005:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-5221:latest
+sweb.eval.x86_64.sympy_s_sympy-17022:latest
+sweb.eval.x86_64.django_s_django-15996:latest
+sweb.eval.x86_64.django_s_django-15252:latest
+sweb.eval.x86_64.sympy_s_sympy-21171:latest
+sweb.eval.x86_64.django_s_django-11797:latest
+sweb.eval.x86_64.django_s_django-16046:latest
+sweb.eval.x86_64.django_s_django-11583:latest
+sweb.eval.x86_64.django_s_django-15738:latest
+sweb.eval.x86_64.sympy_s_sympy-21612:latest
+sweb.eval.x86_64.astropy_s_astropy-12907:latest
+sweb.eval.x86_64.django_s_django-11620:latest
+sweb.eval.x86_64.sympy_s_sympy-16792:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-13779:latest
+sweb.eval.x86_64.django_s_django-16041:latest
+sweb.eval.x86_64.sympy_s_sympy-13471:latest
+sweb.eval.x86_64.sympy_s_sympy-20442:latest
+sweb.eval.x86_64.sympy_s_sympy-20049:latest
+sweb.eval.x86_64.django_s_django-14411:latest
+sweb.eval.x86_64.django_s_django-13447:latest
+sweb.eval.x86_64.django_s_django-12856:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-10949:latest
+sweb.eval.x86_64.django_s_django-14787:latest
+sweb.eval.x86_64.django_s_django-11815:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-13584:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-14087:latest
+sweb.eval.x86_64.django_s_django-15388:latest
+sweb.eval.x86_64.django_s_django-11179:latest
+sweb.eval.x86_64.sympy_s_sympy-24102:latest
+sweb.eval.x86_64.sympy_s_sympy-24213:latest
+sweb.eval.x86_64.django_s_django-15781:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-8906:latest
+sweb.eval.x86_64.django_s_django-13710:latest
+sweb.eval.x86_64.django_s_django-13925:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-14092:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-7373:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-25498:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-5227:latest
+sweb.eval.x86_64.sympy_s_sympy-15678:latest
+sweb.eval.x86_64.django_s_django-13551:latest
+sweb.eval.x86_64.django_s_django-14155:latest
+sweb.eval.x86_64.django_s_django-13933:latest
+sweb.eval.x86_64.sympy_s_sympy-21055:latest
+sweb.eval.x86_64.django_s_django-13660:latest
+sweb.eval.x86_64.django_s_django-16527:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-5692:latest
+sweb.eval.x86_64.mwaskom_s_seaborn-3010:latest
+sweb.eval.x86_64.django_s_django-12700:latest
+sweb.eval.x86_64.sympy_s_sympy-11400:latest
+sweb.eval.x86_64.sympy_s_sympy-23117:latest
+sweb.eval.x86_64.sympy_s_sympy-20639:latest
+sweb.eval.x86_64.sympy_s_sympy-23262:latest
+sweb.eval.x86_64.django_s_django-15498:latest
+sweb.eval.x86_64.django_s_django-12453:latest
+sweb.eval.x86_64.django_s_django-14999:latest
+sweb.eval.x86_64.sympy_s_sympy-13480:latest
+sweb.eval.x86_64.sympy_s_sympy-21847:latest
+sweb.eval.x86_64.sympy_s_sympy-15011:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-25570:latest
+sweb.eval.x86_64.sphinx-doc_s_sphinx-7975:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-14983:latest
+sweb.eval.x86_64.django_s_django-14534:latest
+sweb.eval.x86_64.sympy_s_sympy-14396:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-25442:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-15535:latest
+sweb.eval.x86_64.sympy_s_sympy-22714:latest
+sweb.eval.x86_64.django_s_django-15789:latest
+sweb.eval.x86_64.sympy_s_sympy-21627:latest
+sweb.eval.x86_64.sympy_s_sympy-24066:latest
+sweb.eval.x86_64.pylint-dev_s_pylint-7993:latest
+sweb.eval.x86_64.django_s_django-14752:latest
+sweb.eval.x86_64.sympy_s_sympy-18835:latest
+sweb.eval.x86_64.django_s_django-17051:latest
+sweb.eval.x86_64.sympy_s_sympy-12171:latest
+sweb.eval.x86_64.pydata_s_xarray-3364:latest
+sweb.eval.x86_64.mwaskom_s_seaborn-3190:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-7168:latest
+sweb.eval.x86_64.django_s_django-12747:latest
+sweb.eval.x86_64.django_s_django-15695:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-22835:latest
+sweb.eval.x86_64.sympy_s_sympy-12481:latest
+sweb.eval.x86_64.django_s_django-15851:latest
+sweb.eval.x86_64.sympy_s_sympy-14024:latest
+sweb.eval.x86_64.django_s_django-14608:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-9359:latest
+sweb.eval.x86_64.django_s_django-16873:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-25433:latest
+sweb.eval.x86_64.sympy_s_sympy-13031:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-7432:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-25747:latest
+sweb.eval.x86_64.django_s_django-12286:latest
+sweb.eval.x86_64.django_s_django-11910:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-12471:latest
+sweb.eval.x86_64.pylint-dev_s_pylint-5859:latest
+sweb.eval.x86_64.django_s_django-11133:latest
+sweb.eval.x86_64.astropy_s_astropy-14365:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-13496:latest
+sweb.eval.x86_64.sympy_s_sympy-19487:latest
+sweb.eval.x86_64.sympy_s_sympy-13895:latest
+sweb.eval.x86_64.sympy_s_sympy-15345:latest
+sweb.eval.x86_64.django_s_django-13590:latest
+sweb.eval.x86_64.django_s_django-13757:latest
+sweb.eval.x86_64.django_s_django-16379:latest
+sweb.eval.x86_64.django_s_django-13768:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-8365:latest
+sweb.eval.x86_64.django_s_django-14580:latest
+sweb.eval.x86_64.sympy_s_sympy-20154:latest
+sweb.eval.x86_64.sympy_s_sympy-12419:latest
+sweb.eval.x86_64.django_s_django-12125:latest
+sweb.eval.x86_64.sympy_s_sympy-24152:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-15512:latest
+sweb.eval.x86_64.sympy_s_sympy-18621:latest
+sweb.eval.x86_64.pydata_s_xarray-4248:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-11040:latest
+sweb.eval.x86_64.django_s_django-11099:latest
+sweb.eval.x86_64.django_s_django-16816:latest
+sweb.eval.x86_64.django_s_django-13265:latest
+sweb.eval.x86_64.django_s_django-16139:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-10297:latest
+sweb.eval.x86_64.django_s_django-14016:latest
+sweb.eval.x86_64.pallets_s_flask-5063:latest
+sweb.eval.x86_64.astropy_s_astropy-7746:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-24265:latest
+sweb.eval.x86_64.django_s_django-13448:latest
+sweb.eval.x86_64.django_s_django-12908:latest
+sweb.eval.x86_64.sphinx-doc_s_sphinx-8627:latest
+sweb.eval.x86_64.sympy_s_sympy-14317:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-6116:latest
+sweb.eval.x86_64.sympy_s_sympy-23191:latest
+sweb.eval.x86_64.pydata_s_xarray-5131:latest
+sweb.eval.x86_64.django_s_django-11019:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-23913:latest
+sweb.eval.x86_64.django_s_django-15790:latest
+sweb.eval.x86_64.django_s_django-12497:latest
+sweb.eval.x86_64.matplotlib_s_matplotlib-26020:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-25638:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-25500:latest
+sweb.eval.x86_64.sympy_s_sympy-19007:latest
+sweb.eval.x86_64.django_s_django-12308:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-7220:latest
+sweb.eval.x86_64.django_s_django-11848:latest
+sweb.eval.x86_64.django_s_django-15347:latest
+sweb.eval.x86_64.pytest-dev_s_pytest-7490:latest
+sweb.eval.x86_64.sympy_s_sympy-18532:latest
+sweb.eval.x86_64.django_s_django-14997:latest
+sweb.eval.x86_64.sympy_s_sympy-24909:latest
+sweb.eval.x86_64.django_s_django-13220:latest
+sweb.eval.x86_64.sympy_s_sympy-21614:latest
+sweb.eval.x86_64.django_s_django-15902:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-13497:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-13439:latest
+sweb.eval.x86_64.scikit-learn_s_scikit-learn-14894:latest
+sweb.eval.x86_64.django_s_django-12983:latest
@@ -0,0 +1,31 @@
+def print_diff_ignore_order(file1, file2):
+    with open(file1, 'r') as f1, open(file2, 'r') as f2:
+        file1_lines = set(f1.readlines())
+        file2_lines = set(f2.readlines())
+
+    only_in_file1 = file1_lines - file2_lines
+    only_in_file2 = file2_lines - file1_lines
+
+    if only_in_file1:
+        print(f'Lines in {file1} but not in {file2}:')
+        for line in sorted(only_in_file1):
+            print(f'- {line.strip()}')
+
+    # if only_in_file2:
+    #     print(f"Lines in {file2} but not in {file1}:")
+    #     for line in sorted(only_in_file2):
+    #         print(f"+ {line.strip()}")
+
+    if not only_in_file1 and not only_in_file2:
+        print('The files have the same content (ignoring line order).')
+
+
+if __name__ == '__main__':
+    # Usage
+    lite1 = 'all-swebench-lite-instance-images.txt'  # Replace with the path to your first file
+    lite2 = '../../swe_bench/scripts/docker/all-swebench-lite-instance-images.txt'  # Replace with the path to your second file
+    print_diff_ignore_order(lite1, lite2)
+
+    full1 = 'all-swebench-full-instance-images.txt'  # Replace with the path to your first file
+    full2 = '../../swe_bench/scripts/docker/all-swebench-full-instance-images.txt'  # Replace with the path to your second file
+    print_diff_ignore_order(full1, full2)
@@ -0,0 +1,48 @@
+#!/bin/bash
+# Script will delete all repositories and tags in your Docker Hub account
+set -e
+
+# Set username and password from command-line arguments
+UNAME=$1
+UPASS=$2
+
+# Get token to interact with Docker Hub
+TOKEN=$(curl -s -H "Content-Type: application/json" -X POST -d '{"username": "'${UNAME}'", "password": "'${UPASS}'"}' https://hub.docker.com/v2/users/login/ | jq -r .token)
+
+# Ensure token retrieval was successful
+if [[ -z "$TOKEN" ]]; then
+    echo "Failed to obtain authentication token. Please check your credentials."
+    exit 1
+fi
+
+# Get list of repositories for that user account
+echo "Listing repositories in Docker Hub account '${UNAME}':"
+REPO_LIST=$(curl -s -H "Authorization: JWT ${TOKEN}" "https://hub.docker.com/v2/repositories/${UNAME}/?page_size=10000" | jq -r '.results|.[]|.name')
+if [[ -z "$REPO_LIST" ]]; then
+    echo "No repositories found for user '${UNAME}' or failed to fetch repositories."
+    exit 1
+fi
+
+# Loop through each repository and delete its tags and the repository itself
+for rep in ${REPO_LIST}; do
+    echo "Processing repository: ${UNAME}/${rep}"
+
+    # Get all tags for the repository
+    IMAGES=$(curl -s -H "Authorization: JWT ${TOKEN}" "https://hub.docker.com/v2/repositories/${UNAME}/${rep}/tags/?page_size=100")
+    IMAGE_TAGS=$(echo $IMAGES | jq -r '.results|.[]|.name')
+
+    # Delete each tag
+    for tag in ${IMAGE_TAGS}; do
+        echo "Deleting tag: ${UNAME}/${rep}:${tag}"
+        curl -s -X DELETE -H "Authorization: JWT ${TOKEN}" "https://hub.docker.com/v2/repositories/${UNAME}/${rep}/tags/${tag}/"
+    done
+
+    # Delete the repository itself
+    echo "Deleting repository: ${UNAME}/${rep}"
+    curl -s -X DELETE -H "Authorization: JWT ${TOKEN}" "https://hub.docker.com/v2/repositories/${UNAME}/${rep}/" || {
+        echo "Failed to delete repository '${UNAME}/${rep}'. Please check permissions or API limits."
+    }
+    sleep 1
+done
+
+echo "Script execution completed."
@@ -0,0 +1,18 @@
+from datasets import load_dataset
+
+
+def dataset_to_txt(dataset, txt_file, split='test'):
+    with open(txt_file, 'w') as f:
+        for datum in dataset[split]:
+            instance_id = datum['instance_id'].replace('__', '_s_')
+            f.write(f'sweb.eval.x86_64.{instance_id}:latest\n')
+
+
+if __name__ == '__main__':
+    # Load the private dataset
+    dataset = load_dataset('kjain14/testgeneval')
+
+    dataset_lite = load_dataset('kjain14/testgenevallite')
+
+    dataset_to_txt(dataset_lite, 'all-swebench-lite-instance-images.txt', lite=True)
+    dataset_to_txt(dataset, 'all-swebench-full-instance-images.txt')
@@ -0,0 +1,173 @@
+import argparse
+import copy
+import difflib
+import json
+import os
+import traceback
+
+
+def insert_line_in_string(input_string, new_str, insert_line):
+    """
+    Inserts a new line into a string at the specified line number.
+
+    :param input_string: The original string.
+    :param new_str: The string to insert.
+    :param insert_line: The line number at which to insert (1-based index).
+    :return: The modified string.
+    """
+    file_text = input_string.expandtabs()
+    new_str = new_str.expandtabs()
+
+    file_text_lines = file_text.split('\n')
+
+    new_str_lines = new_str.split('\n')
+    new_file_text_lines = (
+        file_text_lines[:insert_line] + new_str_lines + file_text_lines[insert_line:]
+    )
+
+    return '\n'.join(new_file_text_lines)
+
+
+def print_string_diff(original, modified):
+    """
+    Prints the differences between two strings line by line.
+
+    :param original: The original string.
+    :param modified: The modified string.
+    """
+    original_lines = original.splitlines(keepends=True)
+    modified_lines = modified.splitlines(keepends=True)
+
+    diff = difflib.unified_diff(
+        original_lines,
+        modified_lines,
+        fromfile='original',
+        tofile='modified',
+        lineterm='',
+    )
+
+    print(''.join(diff))
+
+
+def parse_json_files(root_dir, output_dir, metadata_objs, preds_objs):
+    final_output = {i: [] for i in range(25)}
+
+    for subdir in sorted(os.listdir(root_dir)):  # Sorting ensures consistent order
+        subdir_path = os.path.join(root_dir, subdir)
+        # subdir_instance = subdir.rsplit('-', 1)[0]
+        metadata = metadata_objs[subdir]
+        orig_test_suite = metadata['test_result']['test_suite']
+
+        if os.path.isdir(subdir_path):  # Check if it's a directory
+            print(f'Processing subdirectory: {subdir}')
+
+            # Now loop through the JSON files in this subdirectory
+            i = 0
+            test_suite = preds_objs[subdir] if subdir in preds_objs else ''
+            for file in sorted(
+                os.listdir(subdir_path)
+            ):  # Sorting ensures consistent order
+                metadata_copy = copy.deepcopy(metadata)
+                if file.endswith('.json'):  # Check for JSON files
+                    file_path = os.path.join(subdir_path, file)
+                    try:
+                        with open(file_path, 'r', encoding='utf-8') as f:
+                            data = json.load(f)  # Load JSON data
+                            try:
+                                tool_calls = data['response']['choices'][0]['message'][
+                                    'tool_calls'
+                                ]
+                                if tool_calls is not None:
+                                    for tool_call in tool_calls:
+                                        tool_call_dict = eval(
+                                            tool_call['function']['arguments']
+                                        )
+
+                                        if (
+                                            tool_call_dict is not None
+                                            and tool_call_dict != {}
+                                        ):
+                                            command = tool_call_dict['command']
+                                            if command == 'create':
+                                                test_suite = tool_call_dict['file_text']
+                                            if (
+                                                command != 'str_replace'
+                                                and command != 'insert'
+                                                and 'coverage' not in command
+                                            ):
+                                                print(command)
+                                            if command == 'insert':
+                                                test_suite_new = insert_line_in_string(
+                                                    test_suite,
+                                                    tool_call_dict['new_str'],
+                                                    tool_call_dict['insert_line'],
+                                                )
+                                                test_suite = test_suite_new
+                                            if command == 'str_replace':
+                                                if (
+                                                    test_suite.count(
+                                                        tool_call_dict['old_str']
+                                                    )
+                                                    == 1
+                                                ):
+                                                    test_suite_new = test_suite.replace(
+                                                        tool_call_dict['old_str'],
+                                                        tool_call_dict['new_str'],
+                                                    )
+                                                else:
+                                                    continue
+                                                test_suite = test_suite_new
+                            except Exception:
+                                print(traceback.format_exc())
+                                continue
+
+                            metadata_copy['test_result']['test_suite'] = test_suite
+                            if i < 25:
+                                final_output[i].append(metadata_copy)
+                                i += 1
+                    except Exception as e:
+                        print(traceback.format_exc())
+                        print(f'  Error loading {file_path}: {e}')
+
+            for j in range(i, 24):
+                final_output[j].append(metadata_copy)
+            metadata_orig = copy.deepcopy(metadata)
+            metadata_orig['test_result']['test_suite'] = orig_test_suite
+            final_output[24].append(metadata_orig)
+
+    for i in range(25):
+        output_file = os.path.join(output_dir, f'output_{i}.jsonl')
+        with open(output_file, 'w') as f:
+            for metadata in final_output[i]:
+                f.write(json.dumps(metadata) + '\n')
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Parse JSON file')
+    parser.add_argument('--root_dir', type=str, help='Root directory', required=True)
+    parser.add_argument(
+        '--output_dir', type=str, help='Output directory', required=True
+    )
+    parser.add_argument(
+        '--starting_preds_file', type=str, help='Starting predictions', default=None
+    )
+    args = parser.parse_args()
+
+    output_file = os.path.join(args.output_dir, 'output.jsonl')
+    metadata_objs = {}
+    with open(output_file, 'r') as f:
+        content = f.readlines()
+        for line in content:
+            metadata = json.loads(line)
+            metadata_objs[metadata['instance_id']] = metadata
+
+    starting_preds_file = args.starting_preds_file
+    preds_objs = {}
+    if starting_preds_file is not None:
+        with open(starting_preds_file, 'r') as f:
+            content = f.readlines()
+            for line in content:
+                pred = json.loads(line)
+                preds_objs[pred['id']] = pred['preds']['full'][0]
+
+    parse_json_files(args.root_dir, args.output_dir, metadata_objs, preds_objs)
@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+import argparse
+
+import pandas as pd
+
+parser = argparse.ArgumentParser(
+    description='Compare two TestGenEval output JSONL files and print the resolved diff'
+)
+parser.add_argument('input_file_1', type=str)
+parser.add_argument('input_file_2', type=str)
+args = parser.parse_args()
+
+df1 = pd.read_json(args.input_file_1, orient='records', lines=True)
+df2 = pd.read_json(args.input_file_2, orient='records', lines=True)
+
+
+# Get the intersection of the ids
+df = pd.merge(df1, df2, on='id', how='inner')
+
+
+def _get_coverage(report):
+    if report is None:
+        return False
+    if isinstance(report, float):
+        return False
+    else:
+        return report.get('test_pass', False)
+
+
+df['test_pass_x'] = df['test_pass_x'].apply(_get_coverage)
+df['test_pass_y'] = df['test_pass_y'].apply(_get_coverage)
+df['diff'] = df.apply(lambda x: x['test_pass_x'] != x['test_pass_y'], axis=1)
+
+df_diff = df[df['diff']].sort_values(
+    by=['test_pass_x', 'test_pass_y'], ascending=[False, False]
+)
+# skip if any of the pass is nan, which means one of the eval is not finished yet
+df_diff = df_diff[df_diff['test_pass_x'].notna() & df_diff['test_pass_y'].notna()]
+
+print(f'X={args.input_file_1}')
+print(f'Y={args.input_file_2}')
+print(f'# diff={df_diff.shape[0]}')
+df_diff = df_diff[['id', 'test_pass_x', 'test_pass_y', 'report_x', 'report_y']]
+
+# x pass but y not
+print('-' * 100)
+df_diff_x_only = df_diff[df_diff['test_pass_x'] & ~df_diff['test_pass_y']].sort_values(
+    by='id'
+)
+print(f'# x pass but y not={df_diff_x_only.shape[0]}')
+print(df_diff_x_only[['id', 'report_x', 'report_y']])
+
+# y pass but x not
+print('-' * 100)
+df_diff_y_only = df_diff[~df_diff['test_pass_x'] & df_diff['test_pass_y']].sort_values(
+    by='id'
+)
+print(f'# y pass but x not={df_diff_y_only.shape[0]}')
+print(df_diff_y_only[['id', 'report_x', 'report_y']])
+# get instance_id from df_diff_y_only
+print('-' * 100)
+print('Instances that x pass but y not:')
+print(df_diff_x_only['id'].tolist())
+
+print('-' * 100)
+print('Instances that y pass but x not:')
+print(df_diff_y_only['id'].tolist())
@@ -0,0 +1,28 @@
+#!/bin/bash
+
+FOLDER_PATH=$1
+NEW_FOLDER_PATH=${FOLDER_PATH}.swebench_submission
+mkdir -p $NEW_FOLDER_PATH
+
+# Build all_preds.jsonl
+poetry run python evaluation/testgeneval/scripts/eval/convert_oh_output_to_swe_json.py $FOLDER_PATH/output.jsonl
+mv $FOLDER_PATH/output.swebench.jsonl $NEW_FOLDER_PATH/all_preds.jsonl
+
+# Build trajs/
+mkdir -p $NEW_FOLDER_PATH/trajs
+for instance_dir in $FOLDER_PATH/llm_completions/*/; do
+    instance_id=$(basename "$instance_dir")
+    latest_json=$(ls -t "$instance_dir"/*.json | head -n1)
+    if [ -n "$latest_json" ]; then
+        cat "$latest_json" | jq -r '.messages' > "$NEW_FOLDER_PATH/trajs/$instance_id.json"
+    fi
+done
+
+# Build logs/
+# check if $FOLDER_PATH/eval_outputs exists, if so copy over - else raise error
+if [ -d "$FOLDER_PATH/eval_outputs" ]; then
+    cp -r $FOLDER_PATH/eval_outputs $NEW_FOLDER_PATH/logs
+else
+    echo "Error: $FOLDER_PATH/eval_outputs does not exist. You should run the local docker eval_infer.sh first."
+    exit 1
+fi
@@ -0,0 +1,91 @@
+#!/usr/bin/env python3
+"""Convert OpenHands output to a readable markdown format for visualization."""
+
+import argparse
+import json
+import os
+
+import pandas as pd
+from tqdm import tqdm
+
+from evaluation.testgeneval.eval_infer import process_test_suite
+from openhands.events.serialization import event_from_dict
+
+tqdm.pandas()
+
+parser = argparse.ArgumentParser()
+parser.add_argument('oh_output_file', type=str)
+args = parser.parse_args()
+output_md_folder = args.oh_output_file.replace('.jsonl', '.viz')
+print(f'Converting {args.oh_output_file} to markdown files in {output_md_folder}')
+
+oh_format = pd.read_json(args.oh_output_file, orient='records', lines=True)
+# model name is the folder name of oh_output_file
+model_name = os.path.basename(os.path.dirname(args.oh_output_file))
+
+
+def convert_history_to_str(history):
+    ret = ''
+    separator = '\n\n' + '-' * 100 + '\n'
+
+    for i, event in enumerate(history):
+        if i != 0:
+            ret += separator
+
+        if isinstance(event, list):
+            # "event" is a legacy pair of (action, observation)
+            event_obj = event_from_dict(event[0])
+            ret += f'## {i+1}| {event_obj.__class__.__name__}\n\n'
+            ret += str(event_obj)
+            ret += separator
+
+            event_obj = event_from_dict(event[1])
+            ret += f'## {i+1}| {event_obj.__class__.__name__}\n\n'
+            ret += str(event_obj)
+        else:
+            # "event" is a single event
+            event_obj = event_from_dict(event)
+            ret += f'## {i+1}| {event_obj.__class__.__name__}\n\n'
+            ret += str(event_obj)
+    return ret
+
+
+def write_row_to_md_file(row):
+    if 'test_suite' in row:
+        test_suite = row['test_suite']
+    elif 'test_result' in row and 'test_suite' in row['test_result']:
+        test_suite = row['test_result']['test_suite']
+    else:
+        raise ValueError(f'Row {row} does not have a test_suite')
+
+    if 'report' in row:
+        coverage = row['report'].get('coverage', 0)
+        mutation = row['report'].get('mutation_score', 0)
+    else:
+        coverage = None
+        mutation = None
+
+    id = row['id']
+    filename = f'{id}.md'
+    os.makedirs(output_md_folder, exist_ok=True)
+    filepath = os.path.join(output_md_folder, filename)
+
+    with open(filepath, 'w') as f:
+        f.write(f'# {id} (coverage: {coverage})\n')
+        f.write(f'# {id} (mutation score: {mutation})\n')
+
+        # MetaData
+        f.write('## MetaData\n')
+        f.write('```json\n')
+        f.write(json.dumps(row['metadata'], indent=2))
+        f.write('\n```\n')
+
+        # Trajectory
+        f.write('## History\n')
+        f.write(convert_history_to_str(row['history']))
+
+        f.write('## Test Suite\n')
+        f.write(f'{test_suite}\n')
+
+
+oh_format.progress_apply(write_row_to_md_file, axis=1)
@@ -0,0 +1,35 @@
+import argparse
+import os
+
+import pandas as pd
+
+from evaluation.swe_bench.eval_infer import process_git_patch
+
+parser = argparse.ArgumentParser()
+parser.add_argument('oh_output_file', type=str)
+args = parser.parse_args()
+output_filepath = args.oh_output_file.replace('.jsonl', '.swebench.jsonl')
+print(f'Converting {args.oh_output_file} to {output_filepath}')
+
+oh_format = pd.read_json(args.oh_output_file, orient='records', lines=True)
+# model name is the folder name of oh_output_file
+model_name = os.path.basename(os.path.dirname(args.oh_output_file))
+
+
+def convert_row_to_swebench_format(row):
+    if 'git_patch' in row:
+        model_patch = row['git_patch']
+    elif 'test_result' in row and 'git_patch' in row['test_result']:
+        model_patch = row['test_result']['git_patch']
+    else:
+        raise ValueError(f'Row {row} does not have a git_patch')
+
+    return {
+        'instance_id': row['instance_id'],
+        'model_patch': process_git_patch(model_patch),
+        'model_name_or_path': model_name,
+    }
+
+
+swebench_format = oh_format.apply(convert_row_to_swebench_format, axis=1)
+swebench_format.to_json(output_filepath, lines=True, orient='records')
@@ -0,0 +1,27 @@
+import argparse
+
+import pandas as pd
+from datasets import load_dataset
+
+parser = argparse.ArgumentParser()
+parser.add_argument('output_filepath', type=str, help='Path to save the output file')
+parser.add_argument(
+    '--dataset_name',
+    type=str,
+    help='Name of the dataset to download',
+    default='kjain14/testgeneval',
+)
+parser.add_argument('--split', type=str, help='Split to download', default='test')
+args = parser.parse_args()
+
+dataset = load_dataset(args.dataset_name, split=args.split)
+output_filepath = args.output_filepath
+print(
+    f'Downloading gold test suites from {args.dataset_name} (split: {args.split}) to {output_filepath}'
+)
+test_suites = [
+    {'instance_id': row['instance_id'], 'test_suite': row['test_src']} for row in dataset
+]
+print(f'{len(test_suites)} test suites loaded')
+pd.DataFrame(test_suites).to_json(output_filepath, lines=True, orient='records')
+print(f'Test suites saved to {output_filepath}')
@@ -0,0 +1,122 @@
+#!/usr/bin/env python3
+import argparse
+import json
+from collections import Counter
+
+from openhands.events.serialization import event_from_dict
+from openhands.events.utils import get_pairs_from_events
+
+ERROR_KEYWORDS = [
+    'Agent encountered an error while processing the last action',
+    'APIError',
+    'Action execution failed',
+]
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('output_file', type=str, help='The file to summarize')
+    args = parser.parse_args()
+
+    with open(args.output_file, 'r') as file:
+        lines = file.readlines()
+
+    num_lines = len(lines)
+    num_error_lines = 0
+    num_agent_stuck_in_loop = 0
+
+    coverage = 0
+    mutation_score = 0
+    num_empty_suite = 0
+
+    error_counter = Counter()
+
+    main_agent_cost = []
+    editor_cost = []
+    num_turns = []
+
+    for line in lines:
+        _d = json.loads(line)
+
+        # Cost
+        costs = _d['metrics'].get('costs', [])
+        _cur_main_agent_cost = 0
+        _cur_editor_cost = 0
+        for cost in costs:
+            if isinstance(cost, float):
+                # backward compatible
+                _cur_main_agent_cost += cost
+            else:
+                if 'draft_editor' in cost['model']:
+                    _cur_editor_cost += cost['cost']
+                else:
+                    _cur_main_agent_cost += cost['cost']
+
+        main_agent_cost.append(_cur_main_agent_cost)
+        editor_cost.append(_cur_editor_cost)
+
+        # Turn status
+        history = _d.get('history', [])
+        events = [event_from_dict(event) for event in history]
+        pairs = get_pairs_from_events(events)
+        num_turns.append(len(pairs))
+
+        # Suite & resolve status
+        suite = _d.get('test_result', {}).get('test_suite', '')
+        if suite == '':
+            num_empty_suite += 1
+            continue
+
+        report = _d.get('report', {}) or {}
+        coverage += report.get('coverage', 0)
+        mutation_score += report.get('mutation_score', 0)
+
+        # Error
+        error = _d.get('error', None)
+
+        if error is not None and isinstance(error, str):
+            agent_stuck_in_loop = 'Agent got stuck in a loop' in error
+            contains_error = bool(error) and not agent_stuck_in_loop
+            if agent_stuck_in_loop:
+                error_counter['Agent got stuck in a loop'] += 1
+                num_agent_stuck_in_loop += 1
+            elif contains_error:
+                error_counter[error] += 1
+            continue
+
+        for keyword in ERROR_KEYWORDS:
+            if keyword in line:
+                error_counter[keyword] += 1
+                num_error_lines += 1
+                break
+
+    # print the error counter (with percentage)
+    print(
+        f'Average coverage for {num_lines} ({coverage / num_lines * 100:.2f}%)'
+    )
+    print(
+        f'Average mutation score for {num_lines} ({mutation_score / num_lines * 100:.2f}%)'
+    )
+
+    print(
+        f'Number of empty suite: {num_empty_suite} / {num_lines} ({num_empty_suite / num_lines * 100:.2f}%)'
+    )
+    print(
+        f'Number of error lines: {num_error_lines} / {num_lines} ({num_error_lines / num_lines * 100:.2f}%)'
+    )
+    print(
+        f'Number of agent stuck in loop: {num_agent_stuck_in_loop} / {num_lines} ({num_agent_stuck_in_loop / num_lines * 100:.2f}%)'
+    )
+    assert len(num_turns) == num_lines
+    assert len(main_agent_cost) == num_lines
+    assert len(editor_cost) == num_lines
+    print('## Statistics')
+    print(f'Avg. num of turns per instance: {sum(num_turns) / num_lines:.2f}')
+    print(f'Avg. agent cost per instance: {sum(main_agent_cost) / num_lines:.2f} USD')
+    print(f'Avg. editor cost per instance: {sum(editor_cost) / num_lines:.2f} USD')
+    print(
+        f'Avg. total cost per instance: {(sum(main_agent_cost) + sum(editor_cost)) / num_lines:.2f} USD'
+    )
+
+    print('## Detailed error breakdown:')
+    for error, count in error_counter.items():
+        print(f'{error}: {count} ({count / num_lines * 100:.2f}%)')
@@ -0,0 +1,53 @@
+#!/bin/bash
+set -eo pipefail
+
+INPUT_FILE=$1
+NUM_WORKERS=$2
+DATASET=$3
+SPLIT=$4
+SKIP_MUTATION=$5
+
+if [ -z "$INPUT_FILE" ]; then
+  echo "INPUT_FILE not specified (should be a path to a jsonl file)"
+  exit 1
+fi
+
+if [ -z "$DATASET" ]; then
+  echo "DATASET not specified, use default kjain14/testgenevallite"
+  DATASET="kjain14/testgenevallite"
+fi
+
+if [ -z "$SPLIT" ]; then
+  echo "SPLIT not specified, use default test"
+  SPLIT="test"
+fi
+
+if [ -z "$NUM_WORKERS" ]; then
+  echo "NUM_WORKERS not specified, use default 1"
+  NUM_WORKERS=1
+fi
+
+echo "... Evaluating on $INPUT_FILE ..."
+
+COMMAND="poetry run python evaluation/benchmarks/testgeneval/eval_infer.py \
+  --eval-num-workers $NUM_WORKERS \
+  --input-file $INPUT_FILE \
+  --dataset $DATASET \
+  --split $SPLIT"
+
+if [ "$SKIP_MUTATION" == "true" ]; then
+  echo "Skipping mutation evaluation"
+  COMMAND="$COMMAND --skip_mutation"
+fi
+
+if [ -n "$EVAL_LIMIT" ]; then
+  echo "EVAL_LIMIT: $EVAL_LIMIT"
+  COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
+fi
+
+echo $COMMAND
+# Run the command
+eval $COMMAND
+
+# update the output with evaluation results
+# poetry run python evaluation/benchmarks/testgeneval/scripts/eval/update_output_with_eval.py $INPUT_FILE
@@ -0,0 +1,122 @@
+#!/bin/bash
+set -eo pipefail
+
+source "evaluation/utils/version_control.sh"
+
+MODEL_CONFIG=$1
+COMMIT_HASH=$2
+AGENT=$3
+EVAL_LIMIT=$4
+MAX_ITER=$5
+NUM_WORKERS=$6
+DATASET=$7
+SPLIT=$8
+N_RUNS=$9
+ZERO_SHOT_PATH=${10}  # New argument for zero-shot path
+
+if [ -z "$NUM_WORKERS" ]; then
+  NUM_WORKERS=1
+  echo "Number of workers not specified, use default $NUM_WORKERS"
+fi
+checkout_eval_branch
+
+if [ -z "$AGENT" ]; then
+  echo "Agent not specified, use default CodeActAgent"
+  AGENT="CodeActAgent"
+fi
+
+if [ -z "$MAX_ITER" ]; then
+  echo "MAX_ITER not specified, use default 100"
+  MAX_ITER=100
+fi
+
+if [ -z "$USE_INSTANCE_IMAGE" ]; then
+  echo "USE_INSTANCE_IMAGE not specified, use default true"
+  USE_INSTANCE_IMAGE=true
+fi
+
+if [ -z "$RUN_WITH_BROWSING" ]; then
+  echo "RUN_WITH_BROWSING not specified, use default false"
+  RUN_WITH_BROWSING=false
+fi
+
+
+if [ -z "$DATASET" ]; then
+  echo "DATASET not specified, use default princeton-nlp/SWE-bench_Lite"
+  DATASET="princeton-nlp/SWE-bench_Lite"
+fi
+
+if [ -z "$SPLIT" ]; then
+  echo "SPLIT not specified, use default test"
+  SPLIT="test"
+fi
+
+export USE_INSTANCE_IMAGE=$USE_INSTANCE_IMAGE
+echo "USE_INSTANCE_IMAGE: $USE_INSTANCE_IMAGE"
+export RUN_WITH_BROWSING=$RUN_WITH_BROWSING
+echo "RUN_WITH_BROWSING: $RUN_WITH_BROWSING"
+
+get_openhands_version
+
+echo "AGENT: $AGENT"
+echo "OPENHANDS_VERSION: $OPENHANDS_VERSION"
+echo "MODEL_CONFIG: $MODEL_CONFIG"
+echo "DATASET: $DATASET"
+echo "SPLIT: $SPLIT"
+
+# Default to NOT use Hint
+if [ -z "$USE_HINT_TEXT" ]; then
+  export USE_HINT_TEXT=false
+fi
+echo "USE_HINT_TEXT: $USE_HINT_TEXT"
+EVAL_NOTE="$OPENHANDS_VERSION"
+# if not using Hint, add -no-hint to the eval note
+if [ "$USE_HINT_TEXT" = false ]; then
+  EVAL_NOTE="$EVAL_NOTE-no-hint"
+fi
+
+if [ "$RUN_WITH_BROWSING" = true ]; then
+  EVAL_NOTE="$EVAL_NOTE-with-browsing"
+fi
+
+if [ -n "$EXP_NAME" ]; then
+  EVAL_NOTE="$EVAL_NOTE-$EXP_NAME"
+fi
+
+function run_eval() {
+  local eval_note=$1
+  COMMAND="poetry run python evaluation/benchmarks/testgeneval/run_infer.py \
+    --agent-cls $AGENT \
+    --llm-config $MODEL_CONFIG \
+    --max-iterations $MAX_ITER \
+    --eval-num-workers $NUM_WORKERS \
+    --eval-note $eval_note \
+    --dataset $DATASET \
+    --split $SPLIT"
+
+  if [ -n "$EVAL_LIMIT" ]; then
+    echo "EVAL_LIMIT: $EVAL_LIMIT"
+    COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
+  fi
+
+  if [ -n "$ZERO_SHOT_PATH" ]; then
+    echo "ZERO_SHOT_PATH: $ZERO_SHOT_PATH"
+    COMMAND="$COMMAND --testfile_start --zero_shot_path $ZERO_SHOT_PATH"
+  fi
+
+  eval $COMMAND
+}
+
+unset SANDBOX_ENV_GITHUB_TOKEN # prevent the agent from using the github token to push
+if [ -z "$N_RUNS" ]; then
+  N_RUNS=1
+  echo "N_RUNS not specified, use default $N_RUNS"
+fi
+
+for i in $(seq 1 $N_RUNS); do
+  current_eval_note="$EVAL_NOTE-run_$i"
+  echo "EVAL_NOTE: $current_eval_note"
+  run_eval $current_eval_note
+done
+
+checkout_original_branch
@@ -0,0 +1,40 @@
+#!/bin/bash
+
+source ~/.bashrc
+SWEUTIL_DIR=/swe_util
+
+# FIXME: Cannot read SWE_INSTANCE_ID from the environment variable
+# SWE_INSTANCE_ID=django__django-11099
+if [ -z "$SWE_INSTANCE_ID" ]; then
+    echo "Error: SWE_INSTANCE_ID is not set." >&2
+    exit 1
+fi
+
+# Read the swe-bench-test-lite.json file and extract the required item based on instance_id
+item=$(jq --arg INSTANCE_ID "$SWE_INSTANCE_ID" '.[] | select(.instance_id == $INSTANCE_ID)' $SWEUTIL_DIR/eval_data/instances/swe-bench-instance.json)
+
+if [[ -z "$item" ]]; then
+  echo "No item found for the provided instance ID."
+  exit 1
+fi
+
+WORKSPACE_NAME=$(echo "$item" | jq -r '(.repo | tostring) + "__" + (.version | tostring) | gsub("/"; "__")')
+
+echo "WORKSPACE_NAME: $WORKSPACE_NAME"
+
+# Clear the workspace
+if [ -d /workspace ]; then
+    rm -rf /workspace/*
+else
+    mkdir /workspace
+fi
+# Copy repo to workspace
+if [ -d /workspace/$WORKSPACE_NAME ]; then
+    rm -rf /workspace/$WORKSPACE_NAME
+fi
+mkdir -p /workspace
+ln -s /testbed /workspace/$WORKSPACE_NAME
+
+# Activate instance-specific environment
+. /opt/miniconda3/etc/profile.d/conda.sh
+conda activate testbed
@@ -0,0 +1,27 @@
+#!/bin/bash
+
+set -e
+EVAL_WORKSPACE="evaluation/swe_bench/eval_workspace"
+mkdir -p $EVAL_WORKSPACE
+
+# 1. Prepare REPO
+echo "==== Prepare SWE-bench repo ===="
+OH_SWE_BENCH_REPO_PATH="https://github.com/All-Hands-AI/SWE-bench.git"
+OH_SWE_BENCH_REPO_BRANCH="eval"
+git clone -b $OH_SWE_BENCH_REPO_BRANCH $OH_SWE_BENCH_REPO_PATH $EVAL_WORKSPACE/OH-SWE-bench
+
+# 2. Prepare DATA
+echo "==== Prepare SWE-bench data ===="
+EVAL_IMAGE=ghcr.io/all-hands-ai/eval-swe-bench:builder_with_conda
+EVAL_WORKSPACE=$(realpath $EVAL_WORKSPACE)
+chmod +x $EVAL_WORKSPACE/OH-SWE-bench/swebench/harness/prepare_data.sh
+if [ -d $EVAL_WORKSPACE/eval_data ]; then
+    rm -r $EVAL_WORKSPACE/eval_data
+fi
+docker run \
+    -v $EVAL_WORKSPACE:/workspace \
+    -w /workspace \
+    -u $(id -u):$(id -g) \
+    -e HF_DATASETS_CACHE="/tmp" \
+    --rm -it $EVAL_IMAGE \
+    bash -c "cd OH-SWE-bench/swebench/harness && /swe_util/miniforge3/bin/conda run -n swe-bench-eval ./prepare_data.sh && mv eval_data /workspace/"
@@ -0,0 +1,96 @@
+#!/bin/bash
+
+set -e
+
+# assert user name is `root`
+if [ "$USER" != "root" ]; then
+    echo "Error: This script is intended to be run by the 'root' user only." >&2
+    exit 1
+fi
+
+source ~/.bashrc
+
+SWEUTIL_DIR=/swe_util
+
+# Create logs directory
+LOG_DIR=/openhands/logs
+mkdir -p $LOG_DIR && chmod 777 $LOG_DIR
+
+# FIXME: Cannot read SWE_INSTANCE_ID from the environment variable
+# SWE_INSTANCE_ID=django__django-11099
+if [ -z "$SWE_INSTANCE_ID" ]; then
+    echo "Error: SWE_INSTANCE_ID is not set." >&2
+    exit 1
+fi
+
+# Read the swe-bench-test-lite.json file and extract the required item based on instance_id
+item=$(jq --arg INSTANCE_ID "$SWE_INSTANCE_ID" '.[] | select(.instance_id == $INSTANCE_ID)' $SWEUTIL_DIR/eval_data/instances/swe-bench-test-lite.json)
+
+if [[ -z "$item" ]]; then
+  echo "No item found for the provided instance ID."
+  exit 1
+fi
+
+CONDA_ENV_NAME=$(echo "$item" | jq -r '.repo + "__" + .version | gsub("/"; "__")')
+
+echo "CONDA_ENV_NAME: $CONDA_ENV_NAME"
+
+SWE_TASK_DIR=/openhands/swe_tasks
+mkdir -p $SWE_TASK_DIR
+# Dump test_patch to /workspace/test.patch
+echo "$item" | jq -r '.test_patch' > $SWE_TASK_DIR/test.patch
+# Dump patch to /workspace/gold.patch
+echo "$item" | jq -r '.patch' > $SWE_TASK_DIR/gold.patch
+# Dump the item to /workspace/instance.json except for the "test_patch" and "patch" fields
+echo "$item" | jq 'del(.test_patch, .patch)' > $SWE_TASK_DIR/instance.json
+
+# Clear the workspace
+rm -rf /workspace/*
+# Copy repo to workspace
+if [ -d /workspace/$CONDA_ENV_NAME ]; then
+    rm -rf /workspace/$CONDA_ENV_NAME
+fi
+cp -r $SWEUTIL_DIR/eval_data/testbeds/$CONDA_ENV_NAME /workspace
+
+# Reset swe-bench testbed and install the repo
+. $SWEUTIL_DIR/miniforge3/etc/profile.d/conda.sh
+conda config --set changeps1 False
+conda config --append channels conda-forge
+conda activate swe-bench-eval
+
+mkdir -p $SWE_TASK_DIR/reset_testbed_temp
+mkdir -p $SWE_TASK_DIR/reset_testbed_log_dir
+SWE_BENCH_DIR=/swe_util/OH-SWE-bench
+output=$(
+    export PYTHONPATH=$SWE_BENCH_DIR && \
+    cd $SWE_BENCH_DIR && \
+    python swebench/harness/reset_swe_env.py \
+    --swe_bench_tasks $SWEUTIL_DIR/eval_data/instances/swe-bench-test.json \
+    --temp_dir $SWE_TASK_DIR/reset_testbed_temp \
+    --testbed /workspace \
+    --conda_path $SWEUTIL_DIR/miniforge3 \
+    --instance_id $SWE_INSTANCE_ID \
+    --log_dir $SWE_TASK_DIR/reset_testbed_log_dir \
+    --timeout 900 \
+    --verbose
+)
+
+REPO_PATH=$(echo "$output" | awk -F': ' '/repo_path:/ {print $2}')
+TEST_CMD=$(echo "$output" | awk -F': ' '/test_cmd:/ {print $2}')
+echo "Repo Path: $REPO_PATH"
+echo "Test Command: $TEST_CMD"
+
+echo "export SWE_BENCH_DIR=\"$SWE_BENCH_DIR\"" >> ~/.bashrc
+echo "export REPO_PATH=\"$REPO_PATH\"" >> ~/.bashrc
+echo "export TEST_CMD=\"$TEST_CMD\"" >> ~/.bashrc
+
+if [[ "$REPO_PATH" == "None" ]]; then
+    echo "Error: Failed to retrieve repository path. Tests may not have passed or output was not as expected." >&2
+    exit 1
+fi
+
+# Activate instance-specific environment
+. $SWEUTIL_DIR/miniforge3/etc/profile.d/conda.sh
+conda activate $CONDA_ENV_NAME
+
+set +e
@@ -0,0 +1,327 @@
+import ast
+import re
+from typing import List, Tuple
+
+from evaluation.benchmarks.testgeneval.constants import TestStatus
+from evaluation.benchmarks.testgeneval.log_parsers import (
+    MAP_REPO_TO_PARSER,
+    parse_log_pytest,
+)
+
+
+def indent_text(text, indent_level):
+    return '\n'.join(
+        ' ' * indent_level + line if line.strip() else line for line in text.split('\n')
+    )
+
+
+def extract_preamble_classes_and_functions(code):
+    class_pattern = re.compile(
+        r'(?P<decorators>(?:^@[^\r\n]*(?:\r?\n(?:[ \t]+[^\r\n]*|^\)[^\r\n]*)*)*\r?\n)*?)'
+        r'^class\s+([\w]+)(?:\([^)]*\))?:',  # the class line
+        re.MULTILINE,
+    )
+    # Capture methods with or without decorators
+    method_pattern = re.compile(r'(^(\s*@.*\s*)*^\s*def\s+[\w_]+\(.*\):)', re.MULTILINE)
+
+    # Capture functions with or without decorators
+    function_pattern = re.compile(
+        r'(?P<decorators>(?:^@[^\r\n]*(?:\r?\n(?:[ \t]+[^\r\n]*|^\)[^\r\n]*)*)*\r?\n)*?)'
+        r'^def\s+([\w_]+)\(.*\):',  # the function line
+        re.MULTILINE,
+    )
+
+    preamble = ''
+    classes = []
+    test_functions = []
+
+    current_position = 0
+
+    def extract_class_body(code: str, start_index: int) -> Tuple[str, int]:
+        """
+        Extracts the body of a class from the given code starting from the specified index.
+        Returns the class body and the end index of the class body.
+        """
+        if not code or start_index < 0 or start_index >= len(code):
+            raise ValueError('Invalid code or start index')
+
+        # Split the code into lines
+        lines = code[start_index:].split('\n')
+        class_body_lines = []
+
+        # Find the starting indentation level of the class definition
+        class_start_line = lines[0]
+        start_indent = len(class_start_line) - len(class_start_line.lstrip())
+
+        inside_multiline_comment = False
+        end_index = start_index
+        for i, line in enumerate(lines[1:], start=1):
+            stripped_line = line.strip()
+            current_indent = len(line) - len(line.lstrip())
+
+            # Handle multiline comments or docstrings
+            if stripped_line.startswith('"""') or stripped_line.startswith("'''"):
+                if inside_multiline_comment:
+                    inside_multiline_comment = False
+                else:
+                    inside_multiline_comment = True
+
+            if not inside_multiline_comment:
+                # Stop when we reach a line with less indentation than the class definition
+                if current_indent <= start_indent and stripped_line:
+                    break
+
+            # Add lines that are part of the class body
+            class_body_lines.append(line)
+            # Update the end index to the current line end
+            end_index = start_index + len('\n'.join(lines[: i + 1])) + 1
+
+        return code[start_index:end_index], end_index
+
+    while current_position < len(code):
+        class_match = class_pattern.search(code, current_position)
+        method_match = method_pattern.search(code, current_position)
+
+        if class_match and (
+            not method_match or class_match.start() < method_match.start()
+        ):
+            class_name = class_match.group(0)
+            class_body, end_idx = extract_class_body(code, class_match.end())
+            current_position = end_idx
+
+            methods = []
+            class_prefix = class_name
+            set_prefix = False
+            for method_match in method_pattern.finditer(class_body):
+                method_name = method_match.group()
+                method_start = method_match.start()
+                if not set_prefix:
+                    class_prefix = class_name + class_body[:method_start]
+                    set_prefix = True
+                next_method = method_pattern.search(
+                    class_body, method_start + len(method_name)
+                )
+                method_body = (
+                    class_body[method_start : next_method.start()]
+                    if next_method
+                    else class_body[method_start:]
+                )
+                methods.append((method_name, method_body))
+
+            classes.append((class_prefix, methods, class_match.start()))
+
+        elif method_match:
+            function_name = method_match.group(0)
+            start_idx = method_match.start()
+
+            # Extract the current function's indentation level
+            lines = code[start_idx:].split('\n')
+            current_indent = len(lines[0]) - len(lines[0].lstrip())
+
+            next_function = function_pattern.search(
+                code, start_idx + len(function_name)
+            )
+            while next_function and (
+                class_match is None or next_function.start() < class_match.start()
+            ):
+                # Calculate the indentation of the next function
+                next_function_start = next_function.start()
+                next_line = code[next_function_start:].split('\n', 1)[0]
+                next_indent = len(next_line) - len(next_line.lstrip())
+
+                # Check if the next function is top-level
+                if next_indent <= current_indent:
+                    break
+
+                # Continue searching for the next top-level function
+                next_function = function_pattern.search(
+                    code, next_function.start() + len(next_function.group(0))
+                )
+
+            if next_function:
+                next_function_start = next_function.start()
+                if class_match and next_function_start > class_match.start():
+                    next_function_start = class_match.start()
+                function_body = code[start_idx:next_function_start]
+            else:
+                function_body = code[start_idx:]
+
+            test_functions.append((function_body, start_idx))
+            current_position = start_idx + len(function_body)
+
+        else:
+            break
+
+    if classes and test_functions:
+        preamble = code[: min(classes[0][2], test_functions[0][1])]
+    else:
+        preamble = (
+            code[: classes[0][2]]
+            if classes
+            else code[: test_functions[0][1]]
+            if test_functions
+            else code
+        )
+
+    return preamble.strip(), classes, test_functions
+
+
+def filter_passing_tests(
+    test_content: str, test_output: str, repo: str
+) -> Tuple[str, List[str], List[str]]:
+    """
+    Filter tests based on their execution results.
+    Returns:
+        Tuple containing:
+        - Modified test content with only passing tests
+        - List of passing test names
+        - List of failing test names
+    """
+    # Parse test results using appropriate parser
+    parser = MAP_REPO_TO_PARSER.get(repo, parse_log_pytest)
+    test_results = parser(test_output)
+    # Get passing and failing tests
+    passing_tests = []
+    failing_tests = []
+    for test_name, status in test_results.items():
+        if status == TestStatus.PASSED.value:
+            passing_tests.append(test_name)
+        else:
+            failing_tests.append(test_name)
+
+    if not passing_tests:
+        return '', passing_tests, failing_tests
+
+    # Extract test components
+    preamble, classes, functions = extract_preamble_classes_and_functions(test_content)
+
+    # Filter classes to only include passing methods
+    filtered_classes = []
+    for class_name, methods, start_idx in classes:
+        non_fail_methods = []
+        for method_name, method_body in methods:
+            # Extract the base method name for matching
+            method_full_name = (
+                method_name.split('.')[-1].split('(')[0].strip().split(' ')[-1]
+            )
+            # Check if the method name is in failing_tests or if any failing_test is in the method name
+            if not (
+                any(method_full_name in failing_test for failing_test in failing_tests)
+                or any(
+                    failing_test in method_full_name for failing_test in failing_tests
+                )
+            ):
+                non_fail_methods.append((method_name, method_body))
+
+        if non_fail_methods:
+            filtered_classes.append((class_name, non_fail_methods, start_idx))
+
+    # Filter standalone functions
+    filtered_functions = []
+    for func_body, start_idx in functions:
+        func_name = func_body.split('def ')[1].split('(')[0].strip()
+        if any(func_name in failing_test for failing_test in failing_tests) or any(
+            failing_test in func_name for failing_test in failing_tests
+        ):
+            continue
+
+        filtered_functions.append((func_body, start_idx))
+
+    # Reconstruct test content with only passing tests
+    content_parts = [preamble]
+
+    # Add filtered classes
+    for class_name, methods, _ in filtered_classes:
+        class_content = class_name + '\n'
+        for _, method_body in methods:
+            class_content += method_body + '\n'
+        content_parts.append(class_content)
+
+    # Add filtered functions
+    for func_body, _ in filtered_functions:
+        content_parts.append(func_body)
+
+    return '\n\n'.join(content_parts), passing_tests, failing_tests
+
+
+def filter_tests(
+    test_content: str, test_output: str, repo: str
+) -> Tuple[str, List[str], List[str]]:
+    """
+    Filter tests using AST parsing to remove failing test functions from the test file.
+    Non-test functions (e.g. setup or helper methods) and classes (even if all test methods are failing)
+    are preserved.
+
+    If AST processing fails (for example, because the test file cannot be parsed),
+    this function falls back on the existing regex-based filtering (filter_passing_tests).
+
+    Returns:
+        Tuple containing:
+         - Modified test content (as a string) containing only passing tests.
+         - List of passing test names.
+         - List of failing test names.
+    """
+    try:
+        # Attempt to parse the test file using the AST.
+        tree = ast.parse(test_content)
+
+        # Parse test results using the appropriate parser.
+        parser = MAP_REPO_TO_PARSER.get(repo, parse_log_pytest)
+        test_results = parser(test_output)
+        passing_tests = [
+            name
+            for name, status in test_results.items()
+            if status == TestStatus.PASSED.value
+        ]
+        failing_tests = [
+            name
+            for name, status in test_results.items()
+            if status != TestStatus.PASSED.value
+        ]
+
+        # Helper function to decide if a test name should be considered failing.
+        def is_failing(name: str) -> bool:
+            for ft in failing_tests:
+                if name in ft or ft in name:
+                    return True
+            return False
+
+        new_body = []
+        for node in tree.body:
+            # For top-level function definitions, only filter those that look like tests.
+            if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
+                if node.name.startswith('test') and is_failing(node.name):
+                    continue
+                new_body.append(node)
+            # For classes, filter out failing test methods but preserve other methods (e.g. setup).
+            elif isinstance(node, ast.ClassDef):
+                new_class_body = []
+                for subnode in node.body:
+                    if isinstance(subnode, (ast.FunctionDef, ast.AsyncFunctionDef)):
+                        # Only consider filtering if the method is a test.
+                        qualified_name = f'{node.name}.{subnode.name}'
+                        if is_failing(subnode.name) or is_failing(qualified_name):
+                            continue
+                        new_class_body.append(subnode)
+                    else:
+                        new_class_body.append(subnode)
+                # Always include the class even if no test methods remain, as it might contain
+                # setup, teardown, or other necessary logic.
+                if new_class_body:
+                    node.body = new_class_body
+                    new_body.append(node)
+
+            else:
+                new_body.append(node)
+
+        tree.body = new_body
+
+        # Reconstruct the source code from the filtered AST.
+        # (Requires Python 3.9+ for ast.unparse; otherwise an exception will trigger the fallback.)
+        new_test_content = ast.unparse(tree)
+        return new_test_content, passing_tests, failing_tests
+
+    except Exception:
+        print('AST processing failed; falling back on regex-based filtering.')
+        # If AST processing fails for any reason, fall back on the original regex-based filtering.
+        return filter_passing_tests(test_content, test_output, repo)
@@ -0,0 +1,166 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from evaluation.benchmarks.testgeneval.constants import (
+    COVERAGE_PREFIX,
+    KEY_INSTANCE_ID,
+    MAP_REPO_VERSION_TO_SPECS,
+    TESTS_FAILED,
+    TESTS_SUFFIX,
+    UPDATE_TOX,
+    TestGenEvalInstance,
+)
+from evaluation.benchmarks.testgeneval.utils import (
+    get_test_directives,
+)
+
+DIFF_MODIFIED_FILE_REGEX = r'--- a/(.*)'
+
+
+@dataclass
+class TestSpec:
+    """
+    A dataclass that represents a test specification for a single instance of SWE-bench.
+    """
+
+    instance_id: str
+    id: str
+    repo: str
+    version: str
+    test_cmd: str
+    code_file: str
+    test_file: str
+    baseline_covs: dict
+    local_imports: list[str]
+    test_script_list: list[str]
+    mutation_script_list: list[str]
+
+    @property
+    def test_script(self):
+        return (
+            '\n'.join(['#!/bin/bash', 'set -uo pipefail'] + self.test_script_list)
+            + '\n'
+        )
+        # Don't exit early because we need to revert tests at the end
+
+    @property
+    def mutation_script(self):
+        return (
+            '\n'.join(['#!/bin/bash', 'set -uo pipefail'] + self.mutation_script_list)
+            + '\n'
+        )
+        # Don't exit early because we need to revert tests at the end
+
+
+def make_test_setup(specs, env_name, repo_directory, includes_tox=False):
+    eval_commands = []
+
+    if includes_tox:
+        eval_commands.append(UPDATE_TOX)
+
+    eval_commands += [
+        'source /opt/miniconda3/bin/activate',
+        f'conda activate {env_name}',
+        f'cd {repo_directory}',
+    ]
+    if 'eval_commands' in specs:
+        eval_commands += specs['eval_commands']
+    eval_commands += [
+        f'git config --global --add safe.directory {repo_directory}',  # for nonroot user
+        f'cd {repo_directory}',
+        # This is just informational, so we have a record
+        'git status',
+        'git show',
+        'source /opt/miniconda3/bin/activate',
+        f'conda activate {env_name}',
+    ]
+    if 'install' in specs:
+        eval_commands.append(specs['install'])
+
+    if includes_tox:
+        eval_commands.append('add_coverage_tox "tox.ini"')
+
+    eval_commands.append('[ -f ".coveragerc" ] && rm ".coveragerc"')
+    return eval_commands
+
+
+def make_test_script_list(test_cmd, specs, env_name, repo_directory):
+    """
+    Runs the tests.
+    """
+
+    includes_tox = 'tox' in test_cmd
+    eval_commands = make_test_setup(specs, env_name, repo_directory, includes_tox)
+    eval_commands += [
+        f'{test_cmd} || {{ echo "{TESTS_FAILED}\n{TESTS_SUFFIX}\n" && exit 1; }}',
+        f'echo "{TESTS_SUFFIX}"\n',
+        'coverage json -o coverage.json',
+        f'echo "{COVERAGE_PREFIX}"\n',
+        'cat coverage.json',
+    ]
+
+    return eval_commands
+
+
+def make_mutation_script_list(specs, env_name, repo_directory, mutation_timeout):
+    """
+    Runs the tests.
+    """
+
+    eval_commands = make_test_setup(specs, env_name, repo_directory)
+    eval_commands += [
+        'cosmic-ray init mutation.toml mutation.sqlite',
+        f'timeout {mutation_timeout}s cosmic-ray exec mutation.toml mutation.sqlite',
+        'cr-report mutation.sqlite',
+        'cr-rate mutation.sqlite  --estimate --confidence 95.0',
+    ]
+    return eval_commands
+
+
+def make_test_spec(
+    instance: TestGenEvalInstance, mutation_timeout: int, buffer: int
+) -> TestSpec:
+    if isinstance(instance, TestSpec):
+        return instance
+    instance_id = instance[KEY_INSTANCE_ID]
+    id = instance['id']
+    repo = instance['repo']
+    version = instance['version']
+    baseline_covs = instance['baseline_covs']
+    code_file = instance['code_file']
+    test_file = instance['test_file']
+    local_imports = instance['local_imports']
+
+    env_name = 'testbed'
+    repo_directory = f'/{env_name}'
+    specs = MAP_REPO_VERSION_TO_SPECS[repo][version]
+
+    test_cmd = ' '.join(
+        [
+            MAP_REPO_VERSION_TO_SPECS[instance['repo']][instance['version']][
+                'test_cmd'
+            ],
+            *get_test_directives(instance),
+        ]
+    )
+
+    test_script_list = make_test_script_list(test_cmd, specs, env_name, repo_directory)
+
+    mutation_script_list = make_mutation_script_list(
+        specs, env_name, repo_directory, mutation_timeout - buffer
+    )
+
+    return TestSpec(
+        instance_id=instance_id,
+        id=id,
+        repo=repo,
+        test_script_list=test_script_list,
+        test_cmd=test_cmd,
+        local_imports=local_imports,
+        mutation_script_list=mutation_script_list,
+        code_file=code_file,
+        test_file=test_file,
+        baseline_covs=baseline_covs,
+        version=version,
+    )
@@ -0,0 +1,73 @@
+import json
+from pathlib import Path
+from typing import cast
+
+from datasets import Dataset, load_dataset
+
+from evaluation.benchmarks.testgeneval.constants import (
+    KEY_INSTANCE_ID,
+    TestGenEvalInstance,
+)
+
+
+def get_test_directives(instance: TestGenEvalInstance) -> list:
+    """
+    Get test directives from the test_patch of a task instance
+
+    Args:
+        instance (dict): task instance
+    Returns:
+        directives (list): List of test directives
+    """
+    # For seq2seq code repos, testing command is fixed
+    if instance['repo'] == 'swe-bench/humaneval':
+        return ['test.py']
+
+    # Get test directives from test patch and remove non-test files
+    directives = [f"/testbed/{instance['test_file']}"]
+
+    # For Django tests, remove extension + "tests/" prefix and convert slashes to dots (module referencing)
+    if instance['repo'] == 'django/django':
+        directives = [instance['test_file']]
+        directives_transformed = []
+        for d in directives:
+            d = d[: -len('.py')] if d.endswith('.py') else d
+            d = d[len('tests/') :] if d.startswith('tests/') else d
+            d = d.replace('/', '.')
+            directives_transformed.append(d)
+        directives = directives_transformed
+
+    return directives
+
+
+def load_testgeneval_dataset(
+    name='kjain14/testgeneval', split='test', ids=None
+) -> list[TestGenEvalInstance]:
+    """
+    Load SWE-bench dataset from Hugging Face Datasets or local .json/.jsonl file
+    """
+    # check that all instance IDs are in the dataset
+    if ids:
+        ids = set(ids)
+    # Load from local .json/.jsonl file
+    if name.endswith('.json') or name.endswith('.jsonl'):
+        dataset = json.loads(Path(name).read_text())
+        dataset_ids = {instance[KEY_INSTANCE_ID] for instance in dataset}
+    else:
+        # Load from Hugging Face Datasets
+        if name.lower() in {'testgeneval'}:
+            name = 'kjain14/testgeneval'
+        elif name.lower() in {'testgeneval-lite', 'testgenevallite', 'lite'}:
+            name = 'kjain14/testgenevallite'
+        dataset = cast(Dataset, load_dataset(name, split=split))
+        dataset_ids = {instance['id'] for instance in dataset}
+    if ids:
+        if ids - dataset_ids:
+            raise ValueError(
+                (
+                    "Some instance IDs not found in dataset!"
+                    f"\nMissing IDs:\n{' '.join(ids - dataset_ids)}"
+                )
+            )
+        dataset = [instance for instance in dataset if instance['id'] in ids]
+    return [cast(TestGenEvalInstance, instance) for instance in dataset]
@@ -34,7 +34,6 @@ from openhands.utils.async_utils import call_async_from_sync

 FAKE_RESPONSES = {
    'CodeActAgent': fake_user_response,
-    'DelegatorAgent': fake_user_response,
    'VisualBrowsingAgent': fake_user_response,
 }

@@ -4,8 +4,10 @@ import userEvent from "@testing-library/user-event";
 import { afterEach, beforeEach, describe, expect, it, test, vi } from "vitest";
 import OpenHands from "#/api/open-hands";
 import { PaymentForm } from "#/components/features/payment/payment-form";
+import * as featureFlags from "#/utils/feature-flags";

 describe("PaymentForm", () => {
+  const billingSettingsSpy = vi.spyOn(featureFlags, "BILLING_SETTINGS");
  const getBalanceSpy = vi.spyOn(OpenHands, "getBalance");
  const createCheckoutSessionSpy = vi.spyOn(OpenHands, "createCheckoutSession");
  const getConfigSpy = vi.spyOn(OpenHands, "getConfig");
@@ -26,6 +28,7 @@ describe("PaymentForm", () => {
      GITHUB_CLIENT_ID: "123",
      POSTHOG_CLIENT_KEY: "456",
    });
+    billingSettingsSpy.mockReturnValue(true);
  });

  afterEach(() => {
@@ -1,6 +1,7 @@
 import { useQuery } from "@tanstack/react-query";
 import { useConfig } from "./use-config";
 import OpenHands from "#/api/open-hands";
+import { BILLING_SETTINGS } from "#/utils/feature-flags";

 export const useBalance = () => {
  const { data: config } = useConfig();
@@ -8,6 +9,6 @@ export const useBalance = () => {
  return useQuery({
    queryKey: ["user", "balance"],
    queryFn: OpenHands.getBalance,
-    enabled: config?.APP_MODE === "saas",
+    enabled: config?.APP_MODE === "saas" && BILLING_SETTINGS(),
  });
 };
@@ -6,7 +6,6 @@ load_dotenv()
 from openhands.agenthub import (  # noqa: E402
    browsing_agent,
    codeact_agent,
-    delegator_agent,
    dummy_agent,
    visualbrowsing_agent,
 )
@@ -15,7 +14,6 @@ from openhands.controller.agent import Agent  # noqa: E402
 __all__ = [
    'Agent',
    'codeact_agent',
-    'delegator_agent',
    'dummy_agent',
    'browsing_agent',
    'visualbrowsing_agent',
@@ -1,8 +1,6 @@
-import json
 import os
 from collections import deque

-import openhands
 import openhands.agenthub.codeact_agent.function_calling as codeact_function_calling
 from openhands.controller.agent import Agent
 from openhands.controller.state.state import State
@@ -72,23 +70,17 @@ class CodeActAgent(Agent):
            codeact_enable_browsing=self.config.codeact_enable_browsing,
            codeact_enable_jupyter=self.config.codeact_enable_jupyter,
            codeact_enable_llm_editor=self.config.codeact_enable_llm_editor,
+            llm=self.llm,
        )
        logger.debug(
-            f'TOOLS loaded for CodeActAgent: {json.dumps(self.tools, indent=2, ensure_ascii=False).replace("\\n", "\n")}'
+            f"TOOLS loaded for CodeActAgent: {', '.join([tool.get('function').get('name') for tool in self.tools])}"
        )
        self.prompt_manager = PromptManager(
-            microagent_dir=os.path.join(
-                os.path.dirname(os.path.dirname(openhands.__file__)),
-                'microagents',
-            )
-            if self.config.enable_prompt_extensions
-            else None,
            prompt_dir=os.path.join(os.path.dirname(__file__), 'prompts'),
-            disabled_microagents=self.config.disabled_microagents,
        )

        # Create a ConversationMemory instance
-        self.conversation_memory = ConversationMemory(self.prompt_manager)
+        self.conversation_memory = ConversationMemory(self.config, self.prompt_manager)

        self.condenser = Condenser.from_config(self.config.condenser)
        logger.debug(f'Using condenser: {type(self.condenser)}')
@@ -168,7 +160,7 @@ class CodeActAgent(Agent):
        if not self.prompt_manager:
            raise Exception('Prompt Manager not instantiated.')

-        # Use conversation_memory to process events instead of calling events_to_messages directly
+        # Use ConversationMemory to process initial messages
        messages = self.conversation_memory.process_initial_messages(
            with_caching=self.llm.is_caching_prompt_active()
        )
@@ -180,12 +172,12 @@ class CodeActAgent(Agent):
            f'Processing {len(events)} events from a total of {len(state.history)} events'
        )

+        # Use ConversationMemory to process events
        messages = self.conversation_memory.process_events(
            condensed_history=events,
            initial_messages=messages,
            max_message_chars=self.llm.config.max_message_chars,
            vision_is_active=self.llm.vision_is_active(),
-            enable_som_visual_browsing=self.config.enable_som_visual_browsing,
        )

        messages = self._enhance_messages(messages)
@@ -216,14 +208,7 @@ class CodeActAgent(Agent):
                # compose the first user message with examples
                self.prompt_manager.add_examples_to_initial_message(msg)

-                # and/or repo/runtime info
-                if self.config.enable_prompt_extensions:
-                    self.prompt_manager.add_info_to_initial_message(msg)
-
-            # enhance the user message with additional context based on keywords matched
-            if msg.role == 'user':
-                self.prompt_manager.enhance_message(msg)
-
+            elif msg.role == 'user':
                # Add double newline between consecutive user messages
                if prev_role == 'user' and len(msg.content) > 0:
                    # Find the first TextContent in the message to add newlines
@@ -12,13 +12,13 @@ from litellm import (

 from openhands.agenthub.codeact_agent.tools import (
    BrowserTool,
-    CmdRunTool,
    FinishTool,
    IPythonTool,
    LLMBasedFileEditTool,
-    StrReplaceEditorTool,
    ThinkTool,
    WebReadTool,
+    create_cmd_run_tool,
+    create_str_replace_editor_tool,
 )
 from openhands.core.exceptions import (
    FunctionCallNotExistsError,
@@ -39,6 +39,7 @@ from openhands.events.action import (
 )
 from openhands.events.event import FileEditSource, FileReadSource
 from openhands.events.tool import ToolCallMetadata
+from openhands.llm import LLM


 def combine_thought(action: Action, thought: str) -> Action:
@@ -80,7 +81,7 @@ def response_to_actions(response: ModelResponse) -> list[Action]:
            # CmdRunTool (Bash)
            # ================================================

-            if tool_call.function.name == CmdRunTool['function']['name']:
+            if tool_call.function.name == create_cmd_run_tool()['function']['name']:
                if 'command' not in arguments:
                    raise FunctionCallValidationError(
                        f'Missing required argument "command" in tool call {tool_call.function.name}'
@@ -131,7 +132,10 @@ def response_to_actions(response: ModelResponse) -> list[Action]:
                    start=arguments.get('start', 1),
                    end=arguments.get('end', -1),
                )
-            elif tool_call.function.name == StrReplaceEditorTool['function']['name']:
+            elif (
+                tool_call.function.name
+                == create_str_replace_editor_tool()['function']['name']
+            ):
                if 'command' not in arguments:
                    raise FunctionCallValidationError(
                        f'Missing required argument "command" in tool call {tool_call.function.name}'
@@ -219,8 +223,22 @@ def get_tools(
    codeact_enable_browsing: bool = False,
    codeact_enable_llm_editor: bool = False,
    codeact_enable_jupyter: bool = False,
+    llm: LLM | None = None,
 ) -> list[ChatCompletionToolParam]:
-    tools = [CmdRunTool, ThinkTool, FinishTool]
+    SIMPLIFIED_TOOL_DESCRIPTION_LLM_SUBSTRS = ['gpt-', 'o3', 'o1']
+
+    use_simplified_tool_desc = False
+    if llm is not None:
+        use_simplified_tool_desc = any(
+            model_substr in llm.config.model
+            for model_substr in SIMPLIFIED_TOOL_DESCRIPTION_LLM_SUBSTRS
+        )
+
+    tools = [
+        create_cmd_run_tool(use_simplified_description=use_simplified_tool_desc),
+        ThinkTool,
+        FinishTool,
+    ]
    if codeact_enable_browsing:
        tools.append(WebReadTool)
        tools.append(BrowserTool)
@@ -229,5 +247,9 @@ def get_tools(
    if codeact_enable_llm_editor:
        tools.append(LLMBasedFileEditTool)
    else:
-        tools.append(StrReplaceEditorTool)
+        tools.append(
+            create_str_replace_editor_tool(
+                use_simplified_description=use_simplified_tool_desc
+            )
+        )
    return tools
@@ -1,6 +1,6 @@
 {% if repository_info %}
 <REPOSITORY_INFO>
-At the user's request, repository {{ repository_info.repo_name }} has been cloned to directory {{ repository_info.repo_directory }}.
+At the user's request, repository {{ repository_info.repo_name }} has been cloned to the current working directory {{ repository_info.repo_directory }}.
 </REPOSITORY_INFO>
 {% endif %}
 {% if repository_instructions -%}
@@ -20,6 +20,8 @@ When starting a web server, use the corresponding ports. You should also
 set any options to allow iframes and CORS requests, and allow the server to
 be accessed from any host (e.g. 0.0.0.0).
 {% endif %}
+{% if runtime_info.additional_agent_instructions %}
 {{ runtime_info.additional_agent_instructions }}
+{% endif %}
 </RUNTIME_INFORMATION>
 {% endif %}
@@ -1,8 +1,8 @@
 {% for agent_info in triggered_agents %}
 <EXTRA_INFO>
-The following information has been included based on a keyword match for "{{ agent_info.trigger_word }}".
+The following information has been included based on a keyword match for "{{ agent_info.trigger }}".
 It may or may not be relevant to the user's request.

-{{ agent_info.agent.content }}
+{{ agent_info.content }}
 </EXTRA_INFO>
 {% endfor %}
@@ -1,19 +1,19 @@
-from .bash import CmdRunTool
+from .bash import create_cmd_run_tool
 from .browser import BrowserTool
 from .finish import FinishTool
 from .ipython import IPythonTool
 from .llm_based_edit import LLMBasedFileEditTool
-from .str_replace_editor import StrReplaceEditorTool
+from .str_replace_editor import create_str_replace_editor_tool
 from .think import ThinkTool
 from .web_read import WebReadTool

 __all__ = [
    'BrowserTool',
-    'CmdRunTool',
+    'create_cmd_run_tool',
    'FinishTool',
    'IPythonTool',
    'LLMBasedFileEditTool',
-    'StrReplaceEditorTool',
+    'create_str_replace_editor_tool',
    'WebReadTool',
    'ThinkTool',
 ]
@@ -1,6 +1,6 @@
 from litellm import ChatCompletionToolParam, ChatCompletionToolParamFunctionChunk

-_BASH_DESCRIPTION = """Execute a bash command in the terminal within a persistent shell session.
+_DETAILED_BASH_DESCRIPTION = """Execute a bash command in the terminal within a persistent shell session.

 ### Command Execution
 * One command at a time: You can only execute one bash command at a time. If you need to run multiple commands sequentially, use `&&` or `;` to chain them together.
@@ -22,25 +22,39 @@ _BASH_DESCRIPTION = """Execute a bash command in the terminal within a persisten
 * Output truncation: If the output exceeds a maximum length, it will be truncated before being returned.
 """

-CmdRunTool = ChatCompletionToolParam(
-    type='function',
-    function=ChatCompletionToolParamFunctionChunk(
-        name='execute_bash',
-        description=_BASH_DESCRIPTION,
-        parameters={
-            'type': 'object',
-            'properties': {
-                'command': {
-                    'type': 'string',
-                    'description': 'The bash command to execute. Can be empty string to view additional logs when previous exit code is `-1`. Can be `C-c` (Ctrl+C) to interrupt the currently running process. Note: You can only execute one bash command at a time. If you need to run multiple commands sequentially, you can use `&&` or `;` to chain them together.',
-                },
-                'is_input': {
-                    'type': 'string',
-                    'description': 'If True, the command is an input to the running process. If False, the command is a bash command to be executed in the terminal. Default is False.',
-                    'enum': ['true', 'false'],
+_SIMPLIFIED_BASH_DESCRIPTION = """Execute a bash command in the terminal.
+* Long running commands: For commands that may run indefinitely, it should be run in the background and the output should be redirected to a file, e.g. command = `python3 app.py > server.log 2>&1 &`.
+* Interact with running process: If a bash command returns exit code `-1`, this means the process is not yet finished. By setting `is_input` to `true`, the assistant can interact with the running process and send empty `command` to retrieve any additional logs, or send additional text (set `command` to the text) to STDIN of the running process, or send command like `C-c` (Ctrl+C), `C-d` (Ctrl+D), `C-z` (Ctrl+Z) to interrupt the process.
+* One command at a time: You can only execute one bash command at a time. If you need to run multiple commands sequentially, you can use `&&` or `;` to chain them together."""
+
+
+def create_cmd_run_tool(
+    use_simplified_description: bool = False,
+) -> ChatCompletionToolParam:
+    description = (
+        _SIMPLIFIED_BASH_DESCRIPTION
+        if use_simplified_description
+        else _DETAILED_BASH_DESCRIPTION
+    )
+    return ChatCompletionToolParam(
+        type='function',
+        function=ChatCompletionToolParamFunctionChunk(
+            name='execute_bash',
+            description=description,
+            parameters={
+                'type': 'object',
+                'properties': {
+                    'command': {
+                        'type': 'string',
+                        'description': 'The bash command to execute. Can be empty string to view additional logs when previous exit code is `-1`. Can be `C-c` (Ctrl+C) to interrupt the currently running process. Note: You can only execute one bash command at a time. If you need to run multiple commands sequentially, you can use `&&` or `;` to chain them together.',
+                    },
+                    'is_input': {
+                        'type': 'string',
+                        'description': 'If True, the command is an input to the running process. If False, the command is a bash command to be executed in the terminal. Default is False.',
+                        'enum': ['true', 'false'],
+                    },
                },
+                'required': ['command'],
            },
-            'required': ['command'],
-        },
-    ),
-)
+        ),
+    )
@@ -1,6 +1,6 @@
 from litellm import ChatCompletionToolParam, ChatCompletionToolParamFunctionChunk

-_STR_REPLACE_EDITOR_DESCRIPTION = """Custom editing tool for viewing, creating and editing files in plain-text format
+_DETAILED_STR_REPLACE_EDITOR_DESCRIPTION = """Custom editing tool for viewing, creating and editing files in plain-text format
 * State is persistent across command calls and discussions with the user
 * If `path` is a file, `view` displays the result of applying `cat -n`. If `path` is a directory, `view` lists non-hidden files and directories up to 2 levels deep
 * The `create` command cannot be used if the specified `path` already exists as a file
@@ -31,46 +31,73 @@ CRITICAL REQUIREMENTS FOR USING THIS TOOL:
 Remember: when making multiple file edits in a row to the same file, you should prefer to send all edits in a single message with multiple calls to this tool, rather than multiple messages with a single call each.
 """

-StrReplaceEditorTool = ChatCompletionToolParam(
-    type='function',
-    function=ChatCompletionToolParamFunctionChunk(
-        name='str_replace_editor',
-        description=_STR_REPLACE_EDITOR_DESCRIPTION,
-        parameters={
-            'type': 'object',
-            'properties': {
-                'command': {
-                    'description': 'The commands to run. Allowed options are: `view`, `create`, `str_replace`, `insert`, `undo_edit`.',
-                    'enum': ['view', 'create', 'str_replace', 'insert', 'undo_edit'],
-                    'type': 'string',
-                },
-                'path': {
-                    'description': 'Absolute path to file or directory, e.g. `/workspace/file.py` or `/workspace`.',
-                    'type': 'string',
-                },
-                'file_text': {
-                    'description': 'Required parameter of `create` command, with the content of the file to be created.',
-                    'type': 'string',
-                },
-                'old_str': {
-                    'description': 'Required parameter of `str_replace` command containing the string in `path` to replace.',
-                    'type': 'string',
-                },
-                'new_str': {
-                    'description': 'Optional parameter of `str_replace` command containing the new string (if not given, no string will be added). Required parameter of `insert` command containing the string to insert.',
-                    'type': 'string',
-                },
-                'insert_line': {
-                    'description': 'Required parameter of `insert` command. The `new_str` will be inserted AFTER the line `insert_line` of `path`.',
-                    'type': 'integer',
-                },
-                'view_range': {
-                    'description': 'Optional parameter of `view` command when `path` points to a file. If none is given, the full file is shown. If provided, the file will be shown in the indicated line number range, e.g. [11, 12] will show lines 11 and 12. Indexing at 1 to start. Setting `[start_line, -1]` shows all lines from `start_line` to the end of the file.',
-                    'items': {'type': 'integer'},
-                    'type': 'array',
+_SIMPLIFIED_STR_REPLACE_EDITOR_DESCRIPTION = """Custom editing tool for viewing, creating and editing files in plain-text format
+* State is persistent across command calls and discussions with the user
+* If `path` is a file, `view` displays the result of applying `cat -n`. If `path` is a directory, `view` lists non-hidden files and directories up to 2 levels deep
+* The `create` command cannot be used if the specified `path` already exists as a file
+* If a `command` generates a long output, it will be truncated and marked with `<response clipped>`
+* The `undo_edit` command will revert the last edit made to the file at `path`
+Notes for using the `str_replace` command:
+* The `old_str` parameter should match EXACTLY one or more consecutive lines from the original file. Be mindful of whitespaces!
+* If the `old_str` parameter is not unique in the file, the replacement will not be performed. Make sure to include enough context in `old_str` to make it unique
+* The `new_str` parameter should contain the edited lines that should replace the `old_str`
+"""
+
+
+def create_str_replace_editor_tool(
+    use_simplified_description: bool = False,
+) -> ChatCompletionToolParam:
+    description = (
+        _SIMPLIFIED_STR_REPLACE_EDITOR_DESCRIPTION
+        if use_simplified_description
+        else _DETAILED_STR_REPLACE_EDITOR_DESCRIPTION
+    )
+    return ChatCompletionToolParam(
+        type='function',
+        function=ChatCompletionToolParamFunctionChunk(
+            name='str_replace_editor',
+            description=description,
+            parameters={
+                'type': 'object',
+                'properties': {
+                    'command': {
+                        'description': 'The commands to run. Allowed options are: `view`, `create`, `str_replace`, `insert`, `undo_edit`.',
+                        'enum': [
+                            'view',
+                            'create',
+                            'str_replace',
+                            'insert',
+                            'undo_edit',
+                        ],
+                        'type': 'string',
+                    },
+                    'path': {
+                        'description': 'Absolute path to file or directory, e.g. `/workspace/file.py` or `/workspace`.',
+                        'type': 'string',
+                    },
+                    'file_text': {
+                        'description': 'Required parameter of `create` command, with the content of the file to be created.',
+                        'type': 'string',
+                    },
+                    'old_str': {
+                        'description': 'Required parameter of `str_replace` command containing the string in `path` to replace.',
+                        'type': 'string',
+                    },
+                    'new_str': {
+                        'description': 'Optional parameter of `str_replace` command containing the new string (if not given, no string will be added). Required parameter of `insert` command containing the string to insert.',
+                        'type': 'string',
+                    },
+                    'insert_line': {
+                        'description': 'Required parameter of `insert` command. The `new_str` will be inserted AFTER the line `insert_line` of `path`.',
+                        'type': 'integer',
+                    },
+                    'view_range': {
+                        'description': 'Optional parameter of `view` command when `path` points to a file. If none is given, the full file is shown. If provided, the file will be shown in the indicated line number range, e.g. [11, 12] will show lines 11 and 12. Indexing at 1 to start. Setting `[start_line, -1]` shows all lines from `start_line` to the end of the file.',
+                        'items': {'type': 'integer'},
+                        'type': 'array',
+                    },
                },
+                'required': ['command', 'path'],
            },
-            'required': ['command', 'path'],
-        },
-    ),
-)
+        ),
+    )
@@ -1,4 +0,0 @@
-from openhands.agenthub.delegator_agent.agent import DelegatorAgent
-from openhands.controller.agent import Agent
-
-Agent.register('DelegatorAgent', DelegatorAgent)
@@ -1,87 +0,0 @@
-from openhands.controller.agent import Agent
-from openhands.controller.state.state import State
-from openhands.core.config import AgentConfig
-from openhands.events.action import Action, AgentDelegateAction, AgentFinishAction
-from openhands.events.observation import AgentDelegateObservation, Observation
-from openhands.llm.llm import LLM
-
-
-class DelegatorAgent(Agent):
-    VERSION = '1.0'
-    """
-    The Delegator Agent is responsible for delegating tasks to other agents based on the current task.
-    """
-
-    current_delegate: str = ''
-
-    def __init__(self, llm: LLM, config: AgentConfig):
-        """Initialize the Delegator Agent with an LLM
-
-        Parameters:
-        - llm (LLM): The llm to be used by this agent
-        """
-        super().__init__(llm, config)
-
-    def step(self, state: State) -> Action:
-        """Checks to see if current step is completed, returns AgentFinishAction if True.
-        Otherwise, delegates the task to the next agent in the pipeline.
-
-        Parameters:
-        - state (State): The current state given the previous actions and observations
-
-        Returns:
-        - AgentFinishAction: If the last state was 'completed', 'verified', or 'abandoned'
-        - AgentDelegateAction: The next agent to delegate the task to
-        """
-        if self.current_delegate == '':
-            self.current_delegate = 'study'
-            task, _ = state.get_current_user_intent()
-            return AgentDelegateAction(
-                agent='StudyRepoForTaskAgent', inputs={'task': task}
-            )
-
-        # last observation in history should be from the delegate
-        last_observation = None
-        for event in reversed(state.history):
-            if isinstance(event, Observation):
-                last_observation = event
-                break
-
-        if not isinstance(last_observation, AgentDelegateObservation):
-            raise Exception('Last observation is not an AgentDelegateObservation')
-
-        goal, _ = state.get_current_user_intent()
-        if self.current_delegate == 'study':
-            self.current_delegate = 'coder'
-            return AgentDelegateAction(
-                agent='CoderAgent',
-                inputs={
-                    'task': goal,
-                    'summary': last_observation.outputs['summary'],
-                },
-            )
-        elif self.current_delegate == 'coder':
-            self.current_delegate = 'verifier'
-            return AgentDelegateAction(
-                agent='VerifierAgent',
-                inputs={
-                    'task': goal,
-                },
-            )
-        elif self.current_delegate == 'verifier':
-            if (
-                'completed' in last_observation.outputs
-                and last_observation.outputs['completed']
-            ):
-                return AgentFinishAction()
-            else:
-                self.current_delegate = 'coder'
-                return AgentDelegateAction(
-                    agent='CoderAgent',
-                    inputs={
-                        'task': goal,
-                        'summary': last_observation.outputs['summary'],
-                    },
-                )
-        else:
-            raise Exception('Invalid delegate state')
@@ -202,6 +202,7 @@ Note:
        tabs = ''
        last_obs = None
        last_action = None
+        set_of_marks = None  # Initialize set_of_marks to None

        if len(state.history) == 1:
            # for visualwebarena, webarena and miniwob++ eval, we need to retrieve the initial observation already in browser env
@@ -217,6 +218,9 @@ Note:
                # agent has responded, task finished.
                return AgentFinishAction(outputs={'content': event.content})
            elif isinstance(event, Observation):
+                # Only process BrowserOutputObservation and skip other observation types
+                if not isinstance(event, BrowserOutputObservation):
+                    continue
                last_obs = event

        if len(prev_actions) >= 1:  # ignore noop()
@@ -29,7 +29,12 @@ from openhands.core.exceptions import (
 from openhands.core.logger import LOG_ALL_EVENTS
 from openhands.core.logger import openhands_logger as logger
 from openhands.core.schema import AgentState
-from openhands.events import EventSource, EventStream, EventStreamSubscriber
+from openhands.events import (
+    EventSource,
+    EventStream,
+    EventStreamSubscriber,
+    RecallType,
+)
 from openhands.events.action import (
    Action,
    ActionConfirmationStatus,
@@ -42,6 +47,7 @@ from openhands.events.action import (
    MessageAction,
    NullAction,
 )
+from openhands.events.action.agent import RecallAction
 from openhands.events.event import Event
 from openhands.events.observation import (
    AgentCondensationObservation,
@@ -89,7 +95,7 @@ class AgentController:
        max_budget_per_task: float | None = None,
        agent_to_llm_config: dict[str, LLMConfig] | None = None,
        agent_configs: dict[str, AgentConfig] | None = None,
-        sid: str = 'default',
+        sid: str | None = None,
        confirmation_mode: bool = False,
        initial_state: State | None = None,
        is_delegate: bool = False,
@@ -116,7 +122,7 @@ class AgentController:
            status_callback: Optional callback function to handle status updates.
            replay_events: A list of logs to replay.
        """
-        self.id = sid
+        self.id = sid or event_stream.sid
        self.agent = agent
        self.headless_mode = headless_mode
        self.is_delegate = is_delegate
@@ -287,8 +293,14 @@ class AgentController:
                return True
            return False
        if isinstance(event, Observation):
-            if isinstance(event, NullObservation) or isinstance(
-                event, AgentStateChangedObservation
+            if (
+                isinstance(event, NullObservation)
+                and event.cause is not None
+                and event.cause > 0
+            ):
+                return True
+            if isinstance(event, AgentStateChangedObservation) or isinstance(
+                event, NullObservation
            ):
                return False
            return True
@@ -388,6 +400,7 @@ class AgentController:
        if observation.llm_metrics is not None:
            self.agent.llm.metrics.merge(observation.llm_metrics)

+        # this happens for runnable actions and microagent actions
        if self._pending_action and self._pending_action.id == observation.cause:
            if self.state.agent_state == AgentState.AWAITING_USER_CONFIRMATION:
                return
@@ -431,6 +444,25 @@ class AgentController:
                    'debug',
                    f'Extended max iterations to {self.state.max_iterations} after user message',
                )
+            # try to retrieve microagents relevant to the user message
+            # set pending_action while we search for information
+
+            # if this is the first user message for this agent, matters for the microagent info type
+            first_user_message = self._first_user_message()
+            is_first_user_message = (
+                action.id == first_user_message.id if first_user_message else False
+            )
+            recall_type = (
+                RecallType.WORKSPACE_CONTEXT
+                if is_first_user_message
+                else RecallType.KNOWLEDGE
+            )
+
+            recall_action = RecallAction(query=action.content, recall_type=recall_type)
+            self._pending_action = recall_action
+            # this is source=USER because the user message is the trigger for the microagent retrieval
+            self.event_stream.add_event(recall_action, EventSource.USER)
+
            if self.get_agent_state() != AgentState.RUNNING:
                await self.set_agent_state_to(AgentState.RUNNING)
        elif action.source == EventSource.AGENT and action.wait_for_response:
@@ -438,6 +470,7 @@ class AgentController:

    def _reset(self) -> None:
        """Resets the agent controller"""
+        # Runnable actions need an Observation
        # make sure there is an Observation with the tool call metadata to be recognized by the agent
        # otherwise the pending action is found in history, but it's incomplete without an obs with tool result
        if self._pending_action and hasattr(self._pending_action, 'tool_call_metadata'):
@@ -459,6 +492,8 @@ class AgentController:
                obs._cause = self._pending_action.id  # type: ignore[attr-defined]
                self.event_stream.add_event(obs, EventSource.AGENT)

+        # NOTE: RecallActions don't need an ErrorObservation upon reset, as long as they have no tool calls
+
        # reset the pending action, this will be called when the agent is STOPPED or ERROR
        self._pending_action = None
        self.agent.reset()
@@ -1146,3 +1181,26 @@ class AgentController:
                result = event.agent_state == AgentState.RUNNING
                return result
        return False
+
+    def _first_user_message(self) -> MessageAction | None:
+        """
+        Get the first user message for this agent.
+
+        For regular agents, this is the first user message from the beginning (start_id=0).
+        For delegate agents, this is the first user message after the delegate's start_id.
+
+        Returns:
+            MessageAction | None: The first user message, or None if no user message found
+        """
+        # Find the first user message from the appropriate starting point
+        user_messages = list(self.event_stream.get_events(start_id=self.state.start_id))
+
+        # Get and return the first user message
+        return next(
+            (
+                e
+                for e in user_messages
+                if isinstance(e, MessageAction) and e.source == EventSource.USER
+            ),
+            None,
+        )
@@ -135,7 +135,7 @@ class StuckDetector:
        # it takes 3 actions and 3 observations to detect a loop
        # check if the last three actions are the same and result in errors

-        if len(last_actions) < 4 or len(last_observations) < 4:
+        if len(last_actions) < 3 or len(last_observations) < 3:
            return False

        # are the last three actions the "same"?
@@ -17,6 +17,7 @@ from openhands.core.schema import AgentState
 from openhands.core.setup import (
    create_agent,
    create_controller,
+    create_memory,
    create_runtime,
    initialize_repository_for_runtime,
 )
@@ -170,13 +171,22 @@ async def main(loop: asyncio.AbstractEventLoop):
    await runtime.connect()

    # Initialize repository if needed
+    repo_directory = None
    if config.sandbox.selected_repo:
-        initialize_repository_for_runtime(
+        repo_directory = initialize_repository_for_runtime(
            runtime,
-            agent=agent,
            selected_repository=config.sandbox.selected_repo,
        )

+    # when memory is created, it will load the microagents from the selected repository
+    memory = create_memory(
+        runtime=runtime,
+        event_stream=event_stream,
+        sid=sid,
+        selected_repository=config.sandbox.selected_repo,
+        repo_directory=repo_directory,
+    )
+
    if initial_user_action:
        # If there's an initial user action, enqueue it and do not prompt again
        event_stream.add_event(initial_user_action, EventSource.USER)
@@ -185,7 +195,7 @@ async def main(loop: asyncio.AbstractEventLoop):
        asyncio.create_task(prompt_for_next_task())

    await run_agent_until_done(
-        controller, runtime, [AgentState.STOPPED, AgentState.ERROR]
+        controller, runtime, memory, [AgentState.STOPPED, AgentState.ERROR]
    )


@@ -15,6 +15,7 @@ class SandboxConfig(BaseModel):
        timeout: The timeout for the default sandbox action execution.
        remote_runtime_init_timeout: The timeout for the remote runtime to start.
        remote_runtime_api_timeout: The timeout for the remote runtime API requests.
+        remote_runtime_enable_retries: Whether to enable retries (on recoverable errors like requests.ConnectionError) for the remote runtime API requests.
        enable_auto_lint: Whether to enable auto-lint.
        use_host_network: Whether to use the host network.
        runtime_binding_address: The binding address for the runtime ports.  It specifies which network interface on the host machine Docker should bind the runtime ports to.
@@ -53,7 +54,7 @@ class SandboxConfig(BaseModel):
    timeout: int = Field(default=120)
    remote_runtime_init_timeout: int = Field(default=180)
    remote_runtime_api_timeout: int = Field(default=10)
-    remote_runtime_enable_retries: bool = Field(default=False)
+    remote_runtime_enable_retries: bool = Field(default=True)
    remote_runtime_class: str | None = Field(
        default=None
    )  # can be "None" (default to gvisor) or "sysbox" (support docker inside runtime + more stable)
@@ -240,7 +240,7 @@ class SensitiveDataFilter(logging.Filter):
            if (
                len(value) > 2
                and value != 'default'
-                and any(s in key_upper for s in ('SECRET', 'KEY', 'CODE', 'TOKEN'))
+                and any(s in key_upper for s in ('SECRET', '_KEY', '_CODE', '_TOKEN'))
            ):
                sensitive_values.append(value)

@@ -3,12 +3,14 @@ import asyncio
 from openhands.controller import AgentController
 from openhands.core.logger import openhands_logger as logger
 from openhands.core.schema import AgentState
+from openhands.memory.memory import Memory
 from openhands.runtime.base import Runtime


 async def run_agent_until_done(
    controller: AgentController,
    runtime: Runtime,
+    memory: Memory,
    end_states: list[AgentState],
 ):
    """
@@ -37,6 +39,7 @@ async def run_agent_until_done(

    runtime.status_callback = status_callback
    controller.status_callback = status_callback
+    memory.status_callback = status_callback

    while controller.state.agent_state not in end_states:
        await asyncio.sleep(1)
@@ -18,6 +18,7 @@ from openhands.core.schema import AgentState
 from openhands.core.setup import (
    create_agent,
    create_controller,
+    create_memory,
    create_runtime,
    generate_sid,
    initialize_repository_for_runtime,
@@ -29,6 +30,7 @@ from openhands.events.event import Event
 from openhands.events.observation import AgentStateChangedObservation
 from openhands.events.serialization import event_from_dict
 from openhands.io import read_input, read_task
+from openhands.memory.memory import Memory
 from openhands.runtime.base import Runtime
 from openhands.utils.async_utils import call_async_from_sync

@@ -51,6 +53,7 @@ async def run_controller(
    exit_on_message: bool = False,
    fake_user_response_fn: FakeUserResponseFunc | None = None,
    headless_mode: bool = True,
+    memory: Memory | None = None,
 ) -> State | None:
    """Main coroutine to run the agent controller with task input flexibility.

@@ -93,6 +96,8 @@ async def run_controller(
    if agent is None:
        agent = create_agent(config)

+    # when the runtime is created, it will be connected and clone the selected repository
+    repo_directory = None
    if runtime is None:
        runtime = create_runtime(
            config,
@@ -105,14 +110,23 @@ async def run_controller(

        # Initialize repository if needed
        if config.sandbox.selected_repo:
-            initialize_repository_for_runtime(
+            repo_directory = initialize_repository_for_runtime(
                runtime,
-                agent=agent,
                selected_repository=config.sandbox.selected_repo,
            )

    event_stream = runtime.event_stream

+    # when memory is created, it will load the microagents from the selected repository
+    if memory is None:
+        memory = create_memory(
+            runtime=runtime,
+            event_stream=event_stream,
+            sid=sid,
+            selected_repository=config.sandbox.selected_repo,
+            repo_directory=repo_directory,
+        )
+
    replay_events: list[Event] | None = None
    if config.replay_trajectory_path:
        logger.info('Trajectory replay is enabled')
@@ -172,7 +186,7 @@ async def run_controller(
    ]

    try:
-        await run_agent_until_done(controller, runtime, end_states)
+        await run_agent_until_done(controller, runtime, memory, end_states)
    except Exception as e:
        logger.error(f'Exception in main loop: {e}')

@@ -82,5 +82,8 @@ class ActionTypeSchema(BaseModel):
    SEND_PR: str = Field(default='send_pr')
    """Send a PR to github."""

+    RECALL: str = Field(default='recall')
+    """Retrieves content from a user workspace, microagent, or other source."""
+

 ActionType = ActionTypeSchema()
@@ -49,5 +49,8 @@ class ObservationTypeSchema(BaseModel):
    CONDENSE: str = Field(default='condense')
    """Result of a condensation operation."""

+    RECALL: str = Field(default='recall')
+    """Result of a recall operation. This can be the workspace context, a microagent, or other types of information."""
+

 ObservationType = ObservationTypeSchema()
@@ -1,7 +1,7 @@
 import hashlib
 import os
 import uuid
-from typing import Tuple, Type
+from typing import Callable, Tuple, Type

 from pydantic import SecretStr

@@ -16,6 +16,7 @@ from openhands.core.logger import openhands_logger as logger
 from openhands.events import EventStream
 from openhands.events.event import Event
 from openhands.llm.llm import LLM
+from openhands.memory.memory import Memory
 from openhands.microagent.microagent import BaseMicroAgent
 from openhands.runtime import get_runtime_cls
 from openhands.runtime.base import Runtime
@@ -83,7 +84,6 @@ def create_runtime(

 def initialize_repository_for_runtime(
    runtime: Runtime,
-    agent: Agent | None = None,
    selected_repository: str | None = None,
    github_token: SecretStr | None = None,
 ) -> str | None:
@@ -91,7 +91,6 @@ def initialize_repository_for_runtime(

    Args:
        runtime: The runtime to initialize the repository for.
-        agent: (optional) The agent to load microagents for.
        selected_repository: (optional) The GitHub repository to use.
        github_token: (optional) The GitHub token to use.

@@ -99,10 +98,10 @@ def initialize_repository_for_runtime(
        The repository directory path if a repository was cloned, None otherwise.
    """
    # clone selected repository if provided
-    repo_directory = None
    github_token = (
        SecretStr(os.environ.get('GITHUB_TOKEN')) if not github_token else github_token
    )
+    repo_directory = None
    if selected_repository and github_token:
        logger.debug(f'Selected repository {selected_repository}.')
        repo_directory = runtime.clone_repo(
@@ -111,16 +110,47 @@ def initialize_repository_for_runtime(
            None,
        )

-    # load microagents from selected repository
-    if agent and agent.prompt_manager and selected_repository and repo_directory:
-        agent.prompt_manager.set_runtime_info(runtime)
+    return repo_directory
+
+
+def create_memory(
+    runtime: Runtime,
+    event_stream: EventStream,
+    sid: str,
+    selected_repository: str | None = None,
+    repo_directory: str | None = None,
+    status_callback: Callable | None = None,
+) -> Memory:
+    """Create a memory for the agent to use.
+
+    Args:
+        runtime: The runtime to use.
+        event_stream: The event stream it will subscribe to.
+        sid: The session id.
+        selected_repository: The repository to clone and start with, if any.
+        repo_directory: The repository directory, if any.
+        status_callback: Optional callback function to handle status updates.
+    """
+    memory = Memory(
+        event_stream=event_stream,
+        sid=sid,
+        status_callback=status_callback,
+    )
+
+    if runtime:
+        # sets available hosts
+        memory.set_runtime_info(runtime)
+
+        # loads microagents from repo/.openhands/microagents
        microagents: list[BaseMicroAgent] = runtime.get_microagents_from_selected_repo(
            selected_repository
        )
-        agent.prompt_manager.load_microagents(microagents)
-        agent.prompt_manager.set_repository_info(selected_repository, repo_directory)
+        memory.load_user_workspace_microagents(microagents)

-    return repo_directory
+        if selected_repository and repo_directory:
+            memory.set_repository_info(selected_repository, repo_directory)
+
+    return memory


 def create_agent(config: AppConfig) -> Agent:
@@ -1,4 +1,4 @@
-from openhands.events.event import Event, EventSource
+from openhands.events.event import Event, EventSource, RecallType
 from openhands.events.stream import EventStream, EventStreamSubscriber

 __all__ = [
@@ -6,4 +6,5 @@ __all__ = [
    'EventSource',
    'EventStream',
    'EventStreamSubscriber',
+    'RecallType',
 ]
@@ -6,6 +6,7 @@ from openhands.events.action.agent import (
    AgentSummarizeAction,
    AgentThinkAction,
    ChangeAgentStateAction,
+    RecallAction,
 )
 from openhands.events.action.browse import BrowseInteractiveAction, BrowseURLAction
 from openhands.events.action.commands import CmdRunAction, IPythonRunCellAction
@@ -35,4 +36,5 @@ __all__ = [
    'MessageAction',
    'ActionConfirmationStatus',
    'AgentThinkAction',
+    'RecallAction',
 ]
@@ -4,6 +4,7 @@ from typing import Any

 from openhands.core.schema import ActionType
 from openhands.events.action.action import Action
+from openhands.events.event import RecallType


@dataclass
@@ -106,3 +107,22 @@ class AgentDelegateAction(Action):
    @property
    def message(self) -> str:
        return f"I'm asking {self.agent} for help with this task."
+
+
+@dataclass
+class RecallAction(Action):
+    """This action is used for retrieving content, e.g., from the global directory or user workspace."""
+
+    recall_type: RecallType
+    query: str = ''
+    thought: str = ''
+    action: str = ActionType.RECALL
+
+    @property
+    def message(self) -> str:
+        return f'Retrieving content for: {self.query[:50]}'
+
+    def __str__(self) -> str:
+        ret = '**RecallAction**\n'
+        ret += f'QUERY: {self.query[:50]}'
+        return ret
@@ -22,6 +22,16 @@ class FileReadSource(str, Enum):
    DEFAULT = 'default'


+class RecallType(str, Enum):
+    """The type of information that can be retrieved from microagents."""
+
+    WORKSPACE_CONTEXT = 'workspace_context'
+    """Workspace context (repo instructions, runtime, etc.)"""
+
+    KNOWLEDGE = 'knowledge'
+    """A knowledge microagent."""
+
+
@dataclass
 class Event:
    INVALID_ID = -1
@@ -1,7 +1,9 @@
+from openhands.events.event import RecallType
 from openhands.events.observation.agent import (
    AgentCondensationObservation,
    AgentStateChangedObservation,
    AgentThinkObservation,
+    RecallObservation,
 )
 from openhands.events.observation.browse import BrowserOutputObservation
 from openhands.events.observation.commands import (
@@ -40,4 +42,6 @@ __all__ = [
    'SuccessObservation',
    'UserRejectObservation',
    'AgentCondensationObservation',
+    'RecallObservation',
+    'RecallType',
 ]
@@ -1,6 +1,7 @@
-from dataclasses import dataclass
+from dataclasses import dataclass, field

 from openhands.core.schema import ObservationType
+from openhands.events.event import RecallType
 from openhands.events.observation.observation import Observation


@@ -40,3 +41,90 @@ class AgentThinkObservation(Observation):
    @property
    def message(self) -> str:
        return self.content
+
+
+@dataclass
+class MicroagentKnowledge:
+    """
+    Represents knowledge from a triggered microagent.
+
+    Attributes:
+        name: The name of the microagent that was triggered
+        trigger: The word that triggered this microagent
+        content: The actual content/knowledge from the microagent
+    """
+
+    name: str
+    trigger: str
+    content: str
+
+
+@dataclass
+class RecallObservation(Observation):
+    """The retrieval of content from a microagent or more microagents."""
+
+    recall_type: RecallType
+    observation: str = ObservationType.RECALL
+
+    # workspace context
+    repo_name: str = ''
+    repo_directory: str = ''
+    repo_instructions: str = ''
+    runtime_hosts: dict[str, int] = field(default_factory=dict)
+    additional_agent_instructions: str = ''
+
+    # knowledge
+    microagent_knowledge: list[MicroagentKnowledge] = field(default_factory=list)
+    """
+    A list of MicroagentKnowledge objects, each containing information from a triggered microagent.
+
+    Example:
+    [
+        MicroagentKnowledge(
+            name="python_best_practices",
+            trigger="python",
+            content="Always use virtual environments for Python projects."
+        ),
+        MicroagentKnowledge(
+            name="git_workflow",
+            trigger="git",
+            content="Create a new branch for each feature or bugfix."
+        )
+    ]
+    """
+
+    @property
+    def message(self) -> str:
+        return (
+            'Added workspace context'
+            if self.recall_type == RecallType.WORKSPACE_CONTEXT
+            else 'Added microagent knowledge'
+        )
+
+    def __str__(self) -> str:
+        # Build a string representation
+        fields = []
+        if self.recall_type == RecallType.WORKSPACE_CONTEXT:
+            fields.extend(
+                [
+                    f'recall_type={self.recall_type}',
+                    f'repo_name={self.repo_name}',
+                    f'repo_instructions={self.repo_instructions[:20]}...',
+                    f'runtime_hosts={self.runtime_hosts}',
+                    f'additional_agent_instructions={self.additional_agent_instructions[:20]}...',
+                ]
+            )
+        else:
+            fields.extend(
+                [
+                    f'recall_type={self.recall_type}',
+                ]
+            )
+        if self.microagent_knowledge:
+            fields.extend(
+                [
+                    f'microagent_knowledge={", ".join([m.name for m in self.microagent_knowledge])}',
+                ]
+            )
+
+        return f'**RecallObservation**\n{", ".join(fields)}'
@@ -8,6 +8,7 @@ from openhands.events.action.agent import (
    AgentRejectAction,
    AgentThinkAction,
    ChangeAgentStateAction,
+    RecallAction,
 )
 from openhands.events.action.browse import BrowseInteractiveAction, BrowseURLAction
 from openhands.events.action.commands import (
@@ -35,6 +36,7 @@ actions = (
    AgentFinishAction,
    AgentRejectAction,
    AgentDelegateAction,
+    RecallAction,
    ChangeAgentStateAction,
    MessageAction,
 )
@@ -1,5 +1,6 @@
 from dataclasses import asdict
 from datetime import datetime
+from enum import Enum

 from pydantic import BaseModel

@@ -102,6 +103,8 @@ def event_to_dict(event: 'Event') -> dict:
                d['timestamp'] = d['timestamp'].isoformat()
        if key == 'source' and 'source' in d:
            d['source'] = d['source'].value
+        if key == 'recall_type' and 'recall_type' in d:
+            d['recall_type'] = d['recall_type'].value
        if key == 'tool_call_metadata' and 'tool_call_metadata' in d:
            d['tool_call_metadata'] = d['tool_call_metadata'].model_dump()
        if key == 'llm_metrics' and 'llm_metrics' in d:
@@ -119,7 +122,11 @@ def event_to_dict(event: 'Event') -> dict:
        # props is a dict whose values can include a complex object like an instance of a BaseModel subclass
        # such as CmdOutputMetadata
        # we serialize it along with the rest
-        d['extras'] = {k: _convert_pydantic_to_dict(v) for k, v in props.items()}
+        # we also handle the Enum conversion for RecallObservation
+        d['extras'] = {
+            k: (v.value if isinstance(v, Enum) else _convert_pydantic_to_dict(v))
+            for k, v in props.items()
+        }
        # Include success field for CmdOutputObservation
        if hasattr(event, 'success'):
            d['success'] = event.success
@@ -1,9 +1,12 @@
 import copy

+from openhands.events.event import RecallType
 from openhands.events.observation.agent import (
    AgentCondensationObservation,
    AgentStateChangedObservation,
    AgentThinkObservation,
+    MicroagentKnowledge,
+    RecallObservation,
 )
 from openhands.events.observation.browse import BrowserOutputObservation
 from openhands.events.observation.commands import (
@@ -40,6 +43,7 @@ observations = (
    UserRejectObservation,
    AgentCondensationObservation,
    AgentThinkObservation,
+    RecallObservation,
 )

 OBSERVATION_TYPE_TO_CLASS = {
@@ -110,4 +114,18 @@ def observation_from_dict(observation: dict) -> Observation:
        else:
            extras['metadata'] = CmdOutputMetadata()

+    if observation_class is RecallObservation:
+        # handle the Enum conversion
+        if 'recall_type' in extras:
+            extras['recall_type'] = RecallType(extras['recall_type'])
+
+        # convert dicts in microagent_knowledge to MicroagentKnowledge objects
+        if 'microagent_knowledge' in extras and isinstance(
+            extras['microagent_knowledge'], list
+        ):
+            extras['microagent_knowledge'] = [
+                MicroagentKnowledge(**item) if isinstance(item, dict) else item
+                for item in extras['microagent_knowledge']
+            ]
+
    return observation_class(content=content, **extras)
@@ -27,6 +27,7 @@ class EventStreamSubscriber(str, Enum):
    RESOLVER = 'openhands_resolver'
    SERVER = 'server'
    RUNTIME = 'runtime'
+    MEMORY = 'memory'
    MAIN = 'main'
    TEST = 'test'

@@ -5,6 +5,7 @@ from typing import Any
 import httpx
 from pydantic import SecretStr

+from openhands.core.logger import openhands_logger as logger
 from openhands.integrations.service_types import (
    AuthenticationError,
    GitService,
@@ -15,7 +16,7 @@ from openhands.integrations.service_types import (
    User,
 )
 from openhands.utils.import_utils import get_impl
-from openhands.core.logger import openhands_logger as logger
+

 class GitHubService(GitService):
    BASE_URL = 'https://api.github.com'
@@ -25,6 +26,7 @@ class GitHubService(GitService):
    def __init__(
        self,
        user_id: str | None = None,
+        external_auth_id: str | None = None,
        external_auth_token: SecretStr | None = None,
        token: SecretStr | None = None,
        external_token_manager: bool = False,
@@ -249,7 +249,8 @@ class LLM(RetryMixin, DebugMixin):

            # if we mocked function calling, and we have tools, convert the response back to function calling format
            if mock_function_calling and mock_fncall_tools is not None:
-                assert len(resp.choices) == 1
+                logger.debug(f'Response choices: {len(resp.choices)}')
+                assert len(resp.choices) >= 1
                non_fncall_response_message = resp.choices[0].message
                fn_call_messages_with_response = (
                    convert_non_fncall_messages_to_fncall_messages(
@@ -1,5 +1,6 @@
 from litellm import ModelResponse

+from openhands.core.config.agent_config import AgentConfig
 from openhands.core.logger import openhands_logger as logger
 from openhands.core.message import ImageContent, Message, TextContent
 from openhands.core.schema import ActionType
@@ -16,7 +17,7 @@ from openhands.events.action import (
    IPythonRunCellAction,
    MessageAction,
 )
-from openhands.events.event import Event
+from openhands.events.event import Event, RecallType
 from openhands.events.observation import (
    AgentCondensationObservation,
    AgentDelegateObservation,
@@ -28,16 +29,21 @@ from openhands.events.observation import (
    IPythonRunCellObservation,
    UserRejectObservation,
 )
+from openhands.events.observation.agent import (
+    MicroagentKnowledge,
+    RecallObservation,
+)
 from openhands.events.observation.error import ErrorObservation
 from openhands.events.observation.observation import Observation
 from openhands.events.serialization.event import truncate_content
-from openhands.utils.prompt import PromptManager
+from openhands.utils.prompt import PromptManager, RepositoryInfo, RuntimeInfo


 class ConversationMemory:
    """Processes event history into a coherent conversation for the agent."""

-    def __init__(self, prompt_manager: PromptManager):
+    def __init__(self, config: AgentConfig, prompt_manager: PromptManager):
+        self.agent_config = config
        self.prompt_manager = prompt_manager

    def process_events(
@@ -46,23 +52,24 @@ class ConversationMemory:
        initial_messages: list[Message],
        max_message_chars: int | None = None,
        vision_is_active: bool = False,
-        enable_som_visual_browsing: bool = False,
    ) -> list[Message]:
        """Process state history into a list of messages for the LLM.

        Ensures that tool call actions are processed correctly in function calling mode.

        Args:
-            state: The state containing the history of events to convert
-            condensed_history: The condensed list of events to process
-            initial_messages: The initial messages to include in the result
+            condensed_history: The condensed history of events to convert
+            initial_messages: The initial messages to include in the conversation
            max_message_chars: The maximum number of characters in the content of an event included
                in the prompt to the LLM. Larger observations are truncated.
            vision_is_active: Whether vision is active in the LLM. If True, image URLs will be included.
-            enable_som_visual_browsing: Whether to enable visual browsing for the SOM model.
        """
+
        events = condensed_history

+        # log visual browsing status
+        logger.debug(f'Visual browsing: {self.agent_config.enable_som_visual_browsing}')
+
        # Process special events first (system prompts, etc.)
        messages = initial_messages

@@ -70,7 +77,7 @@ class ConversationMemory:
        pending_tool_call_action_messages: dict[str, Message] = {}
        tool_call_id_to_message: dict[str, Message] = {}

-        for event in events:
+        for i, event in enumerate(events):
            # create a regular message from an event
            if isinstance(event, Action):
                messages_to_add = self._process_action(
@@ -84,7 +91,9 @@ class ConversationMemory:
                    tool_call_id_to_message=tool_call_id_to_message,
                    max_message_chars=max_message_chars,
                    vision_is_active=vision_is_active,
-                    enable_som_visual_browsing=enable_som_visual_browsing,
+                    enable_som_visual_browsing=self.agent_config.enable_som_visual_browsing,
+                    current_index=i,
+                    events=events,
                )
            else:
                raise ValueError(f'Unknown event type: {type(event)}')
@@ -270,6 +279,8 @@ class ConversationMemory:
        max_message_chars: int | None = None,
        vision_is_active: bool = False,
        enable_som_visual_browsing: bool = False,
+        current_index: int = 0,
+        events: list[Event] | None = None,
    ) -> list[Message]:
        """Converts an observation into a message format that can be sent to the LLM.

@@ -291,6 +302,8 @@ class ConversationMemory:
            max_message_chars: The maximum number of characters in the content of an observation included in the prompt to the LLM
            vision_is_active: Whether vision is active in the LLM. If True, image URLs will be included
            enable_som_visual_browsing: Whether to enable visual browsing for the SOM model
+            current_index: The index of the current event in the events list (for deduplication)
+            events: The list of all events (for deduplication)

        Returns:
            list[Message]: A list containing the formatted message(s) for the observation.
@@ -372,6 +385,119 @@ class ConversationMemory:
        elif isinstance(obs, AgentCondensationObservation):
            text = truncate_content(obs.content, max_message_chars)
            message = Message(role='user', content=[TextContent(text=text)])
+        elif (
+            isinstance(obs, RecallObservation)
+            and self.agent_config.enable_prompt_extensions
+        ):
+            if obs.recall_type == RecallType.WORKSPACE_CONTEXT:
+                # everything is optional, check if they are present
+                if obs.repo_name or obs.repo_directory:
+                    repo_info = RepositoryInfo(
+                        repo_name=obs.repo_name or '',
+                        repo_directory=obs.repo_directory or '',
+                    )
+                else:
+                    repo_info = None
+
+                if obs.runtime_hosts or obs.additional_agent_instructions:
+                    runtime_info = RuntimeInfo(
+                        available_hosts=obs.runtime_hosts,
+                        additional_agent_instructions=obs.additional_agent_instructions,
+                    )
+                else:
+                    runtime_info = None
+
+                repo_instructions = (
+                    obs.repo_instructions if obs.repo_instructions else ''
+                )
+
+                # Have some meaningful content before calling the template
+                has_repo_info = repo_info is not None and (
+                    repo_info.repo_name or repo_info.repo_directory
+                )
+                has_runtime_info = runtime_info is not None and (
+                    runtime_info.available_hosts
+                    or runtime_info.additional_agent_instructions
+                )
+                has_repo_instructions = bool(repo_instructions.strip())
+
+                # Filter and process microagent knowledge
+                filtered_agents = []
+                if obs.microagent_knowledge:
+                    # Exclude disabled microagents
+                    filtered_agents = [
+                        agent
+                        for agent in obs.microagent_knowledge
+                        if agent.name not in self.agent_config.disabled_microagents
+                    ]
+
+                has_microagent_knowledge = bool(filtered_agents)
+
+                # Generate appropriate content based on what is present
+                message_content = []
+
+                # Build the workspace context information
+                if has_repo_info or has_runtime_info or has_repo_instructions:
+                    formatted_workspace_text = (
+                        self.prompt_manager.build_workspace_context(
+                            repository_info=repo_info,
+                            runtime_info=runtime_info,
+                            repo_instructions=repo_instructions,
+                        )
+                    )
+                    message_content.append(TextContent(text=formatted_workspace_text))
+
+                # Add microagent knowledge if present
+                if has_microagent_knowledge:
+                    formatted_microagent_text = (
+                        self.prompt_manager.build_microagent_info(
+                            triggered_agents=filtered_agents,
+                        )
+                    )
+                    message_content.append(TextContent(text=formatted_microagent_text))
+
+                # Return the combined message if we have any content
+                if message_content:
+                    message = Message(role='user', content=message_content)
+                else:
+                    return []
+            elif obs.recall_type == RecallType.KNOWLEDGE:
+                # Use prompt manager to build the microagent info
+                # First, filter out agents that appear in earlier RecallObservations
+                filtered_agents = self._filter_agents_in_microagent_obs(
+                    obs, current_index, events or []
+                )
+
+                # Create and return a message if there is microagent knowledge to include
+                if filtered_agents:
+                    # Exclude disabled microagents
+                    filtered_agents = [
+                        agent
+                        for agent in filtered_agents
+                        if agent.name not in self.agent_config.disabled_microagents
+                    ]
+
+                    # Only proceed if we still have agents after filtering out disabled ones
+                    if filtered_agents:
+                        formatted_text = self.prompt_manager.build_microagent_info(
+                            triggered_agents=filtered_agents,
+                        )
+
+                        return [
+                            Message(
+                                role='user', content=[TextContent(text=formatted_text)]
+                            )
+                        ]
+
+                # Return empty list if no microagents to include or all were disabled
+                return []
+        elif (
+            isinstance(obs, RecallObservation)
+            and not self.agent_config.enable_prompt_extensions
+        ):
+            # If prompt extensions are disabled, we don't add any additional info
+            # TODO: test this
+            return []
        else:
            # If an observation message is not returned, it will cause an error
            # when the LLM tries to return the next message
@@ -404,3 +530,51 @@ class ConversationMemory:
                    -1
                ].cache_prompt = True  # Last item inside the message content
                break
+
+    def _filter_agents_in_microagent_obs(
+        self, obs: RecallObservation, current_index: int, events: list[Event]
+    ) -> list[MicroagentKnowledge]:
+        """Filter out agents that appear in earlier RecallObservations.
+
+        Args:
+            obs: The current RecallObservation to filter
+            current_index: The index of the current event in the events list
+            events: The list of all events
+
+        Returns:
+            list[MicroagentKnowledge]: The filtered list of microagent knowledge
+        """
+        if obs.recall_type != RecallType.KNOWLEDGE:
+            return obs.microagent_knowledge
+
+        # For each agent in the current microagent observation, check if it appears in any earlier microagent observation
+        filtered_agents = []
+        for agent in obs.microagent_knowledge:
+            # Keep this agent if it doesn't appear in any earlier observation
+            # that is, if this is the first microagent observation with this microagent
+            if not self._has_agent_in_earlier_events(agent.name, current_index, events):
+                filtered_agents.append(agent)
+
+        return filtered_agents
+
+    def _has_agent_in_earlier_events(
+        self, agent_name: str, current_index: int, events: list[Event]
+    ) -> bool:
+        """Check if an agent appears in any earlier RecallObservation in the event list.
+
+        Args:
+            agent_name: The name of the agent to look for
+            current_index: The index of the current event in the events list
+            events: The list of all events
+
+        Returns:
+            bool: True if the agent appears in an earlier RecallObservation, False otherwise
+        """
+        for event in events[:current_index]:
+            # Note that this check includes the WORKSPACE_CONTEXT
+            if isinstance(event, RecallObservation):
+                if any(
+                    agent.name == agent_name for agent in event.microagent_knowledge
+                ):
+                    return True
+        return False
@@ -0,0 +1,292 @@
+import asyncio
+import os
+import uuid
+from typing import Callable
+
+import openhands
+from openhands.core.logger import openhands_logger as logger
+from openhands.events.action.agent import RecallAction
+from openhands.events.event import Event, EventSource, RecallType
+from openhands.events.observation.agent import (
+    MicroagentKnowledge,
+    RecallObservation,
+)
+from openhands.events.observation.empty import NullObservation
+from openhands.events.stream import EventStream, EventStreamSubscriber
+from openhands.microagent import (
+    BaseMicroAgent,
+    KnowledgeMicroAgent,
+    RepoMicroAgent,
+    load_microagents_from_dir,
+)
+from openhands.runtime.base import Runtime
+from openhands.utils.prompt import RepositoryInfo, RuntimeInfo
+
+GLOBAL_MICROAGENTS_DIR = os.path.join(
+    os.path.dirname(os.path.dirname(openhands.__file__)),
+    'microagents',
+)
+
+
+class Memory:
+    """
+    Memory is a component that listens to the EventStream for information retrieval actions
+    (a RecallAction) and publishes observations with the content (such as RecallObservation).
+    """
+
+    sid: str
+    event_stream: EventStream
+    status_callback: Callable | None
+    loop: asyncio.AbstractEventLoop | None
+
+    def __init__(
+        self,
+        event_stream: EventStream,
+        sid: str,
+        status_callback: Callable | None = None,
+    ):
+        self.event_stream = event_stream
+        self.sid = sid if sid else str(uuid.uuid4())
+        self.status_callback = status_callback
+        self.loop = None
+
+        self.event_stream.subscribe(
+            EventStreamSubscriber.MEMORY,
+            self.on_event,
+            self.sid,
+        )
+
+        # Additional placeholders to store user workspace microagents
+        self.repo_microagents: dict[str, RepoMicroAgent] = {}
+        self.knowledge_microagents: dict[str, KnowledgeMicroAgent] = {}
+
+        # Store repository / runtime info to send them to the templating later
+        self.repository_info: RepositoryInfo | None = None
+        self.runtime_info: RuntimeInfo | None = None
+
+        # Load global microagents (Knowledge + Repo)
+        # from typically OpenHands/microagents (i.e., the PUBLIC microagents)
+        self._load_global_microagents()
+
+    def on_event(self, event: Event):
+        """Handle an event from the event stream."""
+        asyncio.get_event_loop().run_until_complete(self._on_event(event))
+
+    async def _on_event(self, event: Event):
+        """Handle an event from the event stream asynchronously."""
+        try:
+            if isinstance(event, RecallAction):
+                # if this is a workspace context recall (on first user message)
+                # create and add a RecallObservation
+                # with info about repo, runtime, instructions, etc. including microagent knowledge if any
+                if (
+                    event.source == EventSource.USER
+                    and event.recall_type == RecallType.WORKSPACE_CONTEXT
+                ):
+                    logger.debug('Workspace context recall')
+                    workspace_obs: RecallObservation | NullObservation | None = None
+
+                    workspace_obs = self._on_workspace_context_recall(event)
+                    if workspace_obs is None:
+                        workspace_obs = NullObservation(content='')
+
+                    # important: this will release the execution flow from waiting for the retrieval to complete
+                    workspace_obs._cause = event.id  # type: ignore[union-attr]
+
+                    self.event_stream.add_event(workspace_obs, EventSource.ENVIRONMENT)
+                    return
+
+                # Handle knowledge recall (triggered microagents)
+                elif (
+                    event.source == EventSource.USER
+                    and event.recall_type == RecallType.KNOWLEDGE
+                ):
+                    logger.debug('Microagent knowledge recall')
+                    microagent_obs: RecallObservation | NullObservation | None = None
+                    microagent_obs = self._on_microagent_recall(event)
+                    if microagent_obs is None:
+                        microagent_obs = NullObservation(content='')
+
+                    # important: this will release the execution flow from waiting for the retrieval to complete
+                    microagent_obs._cause = event.id  # type: ignore[union-attr]
+
+                    self.event_stream.add_event(microagent_obs, EventSource.ENVIRONMENT)
+                    return
+        except Exception as e:
+            error_str = f'Error: {str(e.__class__.__name__)}'
+            logger.error(error_str)
+            self.send_error_message('STATUS$ERROR_MEMORY', error_str)
+            return
+
+    def _on_workspace_context_recall(
+        self, event: RecallAction
+    ) -> RecallObservation | None:
+        """Add repository and runtime information to the stream as a RecallObservation."""
+
+        # Create WORKSPACE_CONTEXT info:
+        # - repository_info
+        # - runtime_info
+        # - repository_instructions
+        # - microagent_knowledge
+
+        # Collect raw repository instructions
+        repo_instructions = ''
+        assert (
+            len(self.repo_microagents) <= 1
+        ), f'Expecting at most one repo microagent, but found {len(self.repo_microagents)}: {self.repo_microagents.keys()}'
+
+        # Retrieve the context of repo instructions
+        for microagent in self.repo_microagents.values():
+            if repo_instructions:
+                repo_instructions += '\n\n'
+            repo_instructions += microagent.content
+
+        # Find any matched microagents based on the query
+        microagent_knowledge = self._find_microagent_knowledge(event.query)
+
+        # Create observation if we have anything
+        if (
+            self.repository_info
+            or self.runtime_info
+            or repo_instructions
+            or microagent_knowledge
+        ):
+            obs = RecallObservation(
+                recall_type=RecallType.WORKSPACE_CONTEXT,
+                repo_name=self.repository_info.repo_name
+                if self.repository_info and self.repository_info.repo_name is not None
+                else '',
+                repo_directory=self.repository_info.repo_directory
+                if self.repository_info
+                and self.repository_info.repo_directory is not None
+                else '',
+                repo_instructions=repo_instructions if repo_instructions else '',
+                runtime_hosts=self.runtime_info.available_hosts
+                if self.runtime_info and self.runtime_info.available_hosts is not None
+                else {},
+                additional_agent_instructions=self.runtime_info.additional_agent_instructions
+                if self.runtime_info
+                and self.runtime_info.additional_agent_instructions is not None
+                else '',
+                microagent_knowledge=microagent_knowledge,
+                content='Added workspace context',
+            )
+            return obs
+        return None
+
+    def _on_microagent_recall(
+        self,
+        event: RecallAction,
+    ) -> RecallObservation | None:
+        """When a microagent action triggers microagents, create a RecallObservation with structured data."""
+
+        # Find any matched microagents based on the query
+        microagent_knowledge = self._find_microagent_knowledge(event.query)
+
+        # Create observation if we have anything
+        if microagent_knowledge:
+            obs = RecallObservation(
+                recall_type=RecallType.KNOWLEDGE,
+                microagent_knowledge=microagent_knowledge,
+                content='Retrieved knowledge from microagents',
+            )
+            return obs
+        return None
+
+    def _find_microagent_knowledge(self, query: str) -> list[MicroagentKnowledge]:
+        """Find microagent knowledge based on a query.
+
+        Args:
+            query: The query to search for microagent triggers
+
+        Returns:
+            A list of MicroagentKnowledge objects for matched triggers
+        """
+        recalled_content: list[MicroagentKnowledge] = []
+
+        # skip empty queries
+        if not query:
+            return recalled_content
+
+        # Search for microagent triggers in the query
+        for name, microagent in self.knowledge_microagents.items():
+            trigger = microagent.match_trigger(query)
+            if trigger:
+                logger.info("Microagent '%s' triggered by keyword '%s'", name, trigger)
+                recalled_content.append(
+                    MicroagentKnowledge(
+                        name=microagent.name,
+                        trigger=trigger,
+                        content=microagent.content,
+                    )
+                )
+        return recalled_content
+
+    def load_user_workspace_microagents(
+        self, user_microagents: list[BaseMicroAgent]
+    ) -> None:
+        """
+        This method loads microagents from a user's cloned repo or workspace directory.
+
+        This is typically called from agent_session or setup once the workspace is cloned.
+        """
+        logger.info(
+            'Loading user workspace microagents: %s', [m.name for m in user_microagents]
+        )
+        for user_microagent in user_microagents:
+            if isinstance(user_microagent, KnowledgeMicroAgent):
+                self.knowledge_microagents[user_microagent.name] = user_microagent
+            elif isinstance(user_microagent, RepoMicroAgent):
+                self.repo_microagents[user_microagent.name] = user_microagent
+
+    def _load_global_microagents(self) -> None:
+        """
+        Loads microagents from the global microagents_dir
+        """
+        repo_agents, knowledge_agents, _ = load_microagents_from_dir(
+            GLOBAL_MICROAGENTS_DIR
+        )
+        for name, agent in knowledge_agents.items():
+            if isinstance(agent, KnowledgeMicroAgent):
+                self.knowledge_microagents[name] = agent
+        for name, agent in repo_agents.items():
+            if isinstance(agent, RepoMicroAgent):
+                self.repo_microagents[name] = agent
+
+    def set_repository_info(self, repo_name: str, repo_directory: str) -> None:
+        """Store repository info so we can reference it in an observation."""
+        if repo_name or repo_directory:
+            self.repository_info = RepositoryInfo(repo_name, repo_directory)
+        else:
+            self.repository_info = None
+
+    def set_runtime_info(self, runtime: Runtime) -> None:
+        """Store runtime info (web hosts, ports, etc.)."""
+        # e.g. { '127.0.0.1': 8080 }
+        if runtime.web_hosts or runtime.additional_agent_instructions:
+            self.runtime_info = RuntimeInfo(
+                available_hosts=runtime.web_hosts,
+                additional_agent_instructions=runtime.additional_agent_instructions,
+            )
+        else:
+            self.runtime_info = None
+
+    def send_error_message(self, message_id: str, message: str):
+        """Sends an error message if the callback function was provided."""
+        if self.status_callback:
+            try:
+                if self.loop is None:
+                    self.loop = asyncio.get_running_loop()
+                asyncio.run_coroutine_threadsafe(
+                    self._send_status_message('error', message_id, message), self.loop
+                )
+            except RuntimeError as e:
+                logger.error(
+                    f'Error sending status message: {e.__class__.__name__}',
+                    stack_info=False,
+                )
+
+    async def _send_status_message(self, msg_type: str, id: str, message: str):
+        """Sends a status message to the client."""
+        if self.status_callback:
+            self.status_callback(msg_type, id, message)
@@ -27,7 +27,7 @@ class GithubIssueHandler(IssueHandlerInterface):

    def get_headers(self) -> dict[str, str]:
        return {
-            'Authorization': f'Bearer {self.token}',
+            'Authorization': f'token {self.token}',
            'Accept': 'application/vnd.github.v3+json',
        }

@@ -450,7 +450,7 @@ class GithubPRHandler(GithubIssueHandler):
        """Download comments for a specific pull request from Github."""
        url = f'https://api.github.com/repos/{self.owner}/{self.repo}/issues/{pr_number}/comments'
        headers = {
-            'Authorization': f'Bearer {self.token}',
+            'Authorization': f'token {self.token}',
            'Accept': 'application/vnd.github.v3+json',
        }
        params = {'per_page': 100, 'page': 1}
@@ -97,7 +97,7 @@ class Runtime(FileEditRuntimeMixin):
        status_callback: Callable | None = None,
        attach_to_existing: bool = False,
        headless_mode: bool = False,
-        github_user_id: str | None = None,
+        user_id: str | None = None,
    ):
        self.sid = sid
        self.event_stream = event_stream
@@ -130,7 +130,7 @@ class Runtime(FileEditRuntimeMixin):
            self, enable_llm_editor=config.get_agent_config().codeact_enable_llm_editor
        )

-        self.github_user_id = github_user_id
+        self.user_id = user_id

    def setup_initial_env(self) -> None:
        if self.attach_to_existing:
@@ -220,9 +220,9 @@ class Runtime(FileEditRuntimeMixin):
        assert event.timeout is not None
        try:
            if isinstance(event, CmdRunAction):
-                if self.github_user_id and '$GITHUB_TOKEN' in event.command:
+                if self.user_id and '$GITHUB_TOKEN' in event.command:
                    gh_client = GithubServiceImpl(
-                        user_id=self.github_user_id, external_token_manager=True
+                        external_auth_id=self.user_id, external_token_manager=True
                    )
                    token = await gh_client.get_latest_token()
                    if token:
@@ -59,7 +59,7 @@ class ActionExecutionClient(Runtime):
        status_callback: Any | None = None,
        attach_to_existing: bool = False,
        headless_mode: bool = True,
-        github_user_id: str | None = None,
+        user_id: str | None = None,
    ):
        self.session = HttpSession()
        self.action_semaphore = threading.Semaphore(1)  # Ensure one action at a time
@@ -75,7 +75,7 @@ class ActionExecutionClient(Runtime):
            status_callback,
            attach_to_existing,
            headless_mode,
-            github_user_id,
+            user_id,
        )

    @abstractmethod
@@ -1,3 +1,4 @@
+import logging
 import os
 from typing import Callable
 from urllib.parse import urlparse
@@ -45,7 +46,7 @@ class RemoteRuntime(ActionExecutionClient):
        status_callback: Callable | None = None,
        attach_to_existing: bool = False,
        headless_mode: bool = True,
-        github_user_id: str | None = None,
+        user_id: str | None = None,
    ):
        super().__init__(
            config,
@@ -56,7 +57,7 @@ class RemoteRuntime(ActionExecutionClient):
            status_callback,
            attach_to_existing,
            headless_mode,
-            github_user_id,
+            user_id,
        )
        if self.config.sandbox.api_key is None:
            raise ValueError(
@@ -425,10 +426,11 @@ class RemoteRuntime(ActionExecutionClient):
            return self._send_action_server_request_impl(method, url, **kwargs)

        retry_decorator = tenacity.retry(
-            retry=tenacity.retry_if_exception_type(ConnectionError),
+            retry=tenacity.retry_if_exception_type(requests.ConnectionError),
            stop=tenacity.stop_after_attempt(3)
            | stop_if_should_exit()
            | self._stop_if_closed,
+            before_sleep=tenacity.before_sleep_log(logger, logging.WARNING),
            wait=tenacity.wait_exponential(multiplier=1, min=4, max=60),
        )
        return retry_decorator(self._send_action_server_request_impl)(
@@ -46,7 +46,12 @@ class ConversationManager(ABC):

    @abstractmethod
    async def join_conversation(
-        self, sid: str, connection_id: str, settings: Settings, user_id: str | None
+        self,
+        sid: str,
+        connection_id: str,
+        settings: Settings,
+        user_id: str | None,
+        github_user_id: str | None,
    ) -> EventStream | None:
        """Join a conversation and return its event stream."""

@@ -74,6 +79,7 @@ class ConversationManager(ABC):
        settings: Settings,
        user_id: str | None,
        initial_user_msg: MessageAction | None = None,
+        github_user_id: str | None = None,
    ) -> EventStream:
        """Start an event loop if one is not already running"""

@@ -106,7 +106,12 @@ class StandaloneConversationManager(ConversationManager):
            return c

    async def join_conversation(
-        self, sid: str, connection_id: str, settings: Settings, user_id: str | None
+        self,
+        sid: str,
+        connection_id: str,
+        settings: Settings,
+        user_id: str | None,
+        github_user_id: str | None,
    ):
        logger.info(
            f'join_conversation:{sid}:{connection_id}',
@@ -116,7 +121,9 @@ class StandaloneConversationManager(ConversationManager):
        self._local_connection_id_to_session_id[connection_id] = sid
        event_stream = await self._get_event_stream(sid)
        if not event_stream:
-            return await self.maybe_start_agent_loop(sid, settings, user_id)
+            return await self.maybe_start_agent_loop(
+                sid, settings, user_id, github_user_id=github_user_id
+            )
        for event in event_stream.get_events(reverse=True):
            if isinstance(event, AgentStateChangedObservation):
                if event.agent_state in (
@@ -187,14 +194,18 @@ class StandaloneConversationManager(ConversationManager):
                logger.error('error_cleaning_stale')
                await asyncio.sleep(_CLEANUP_INTERVAL)

-    async def _get_conversation_store(self, user_id: str | None) -> ConversationStore:
+    async def _get_conversation_store(
+        self, user_id: str | None, github_user_id: str | None
+    ) -> ConversationStore:
        conversation_store_class = self._conversation_store_class
        if not conversation_store_class:
            self._conversation_store_class = conversation_store_class = get_impl(
                ConversationStore,  # type: ignore
                self.server_config.conversation_store_class,
            )
-        store = await conversation_store_class.get_instance(self.config, user_id)
+        store = await conversation_store_class.get_instance(
+            self.config, user_id, github_user_id
+        )
        return store

    async def get_running_agent_loops(
@@ -243,6 +254,7 @@ class StandaloneConversationManager(ConversationManager):
        settings: Settings,
        user_id: str | None,
        initial_user_msg: MessageAction | None = None,
+        github_user_id: str | None = None,
    ) -> EventStream:
        logger.info(f'maybe_start_agent_loop:{sid}', extra={'session_id': sid})
        session: Session | None = None
@@ -256,7 +268,9 @@ class StandaloneConversationManager(ConversationManager):
                    extra={'session_id': sid, 'user_id': user_id},
                )
                # Get the conversations sorted (oldest first)
-                conversation_store = await self._get_conversation_store(user_id)
+                conversation_store = await self._get_conversation_store(
+                    user_id, github_user_id
+                )
                conversations = await conversation_store.get_all_metadata(response_ids)
                conversations.sort(key=_last_updated_at_key, reverse=True)

@@ -277,7 +291,9 @@ class StandaloneConversationManager(ConversationManager):
            try:
                session.agent_session.event_stream.subscribe(
                    EventStreamSubscriber.SERVER,
-                    self._create_conversation_update_callback(user_id, sid),
+                    self._create_conversation_update_callback(
+                        user_id, github_user_id, sid
+                    ),
                    UPDATED_AT_CALLBACK_ID,
                )
            except ValueError:
@@ -374,22 +390,23 @@ class StandaloneConversationManager(ConversationManager):
        )

    def _create_conversation_update_callback(
-        self, user_id: str | None, conversation_id: str
+        self, user_id: str | None, github_user_id: str | None, conversation_id: str
    ) -> Callable:
        def callback(*args, **kwargs):
            call_async_from_sync(
                self._update_timestamp_for_conversation,
                GENERAL_TIMEOUT,
                user_id,
+                github_user_id,
                conversation_id,
            )

        return callback

    async def _update_timestamp_for_conversation(
-        self, user_id: str, conversation_id: str
+        self, user_id: str, github_user_id: str, conversation_id: str
    ):
-        conversation_store = await self._get_conversation_store(user_id)
+        conversation_store = await self._get_conversation_store(user_id, github_user_id)
        conversation = await conversation_store.get_metadata(conversation_id)
        conversation.last_updated_at = datetime.now(timezone.utc)
        await conversation_store.save_metadata(conversation)
@@ -6,10 +6,14 @@ from openhands.core.logger import openhands_logger as logger
 from openhands.events.action import (
    NullAction,
 )
+from openhands.events.action.agent import RecallAction
 from openhands.events.observation import (
    NullObservation,
 )
-from openhands.events.observation.agent import AgentStateChangedObservation
+from openhands.events.observation.agent import (
+    AgentStateChangedObservation,
+    RecallObservation,
+)
 from openhands.events.serialization import event_to_dict
 from openhands.events.stream import AsyncEventStreamWrapper
 from openhands.server.shared import (
@@ -35,7 +39,9 @@ async def connect(connection_id: str, environ):

    cookies_str = environ.get('HTTP_COOKIE', '')
    conversation_validator = ConversationValidatorImpl()
-    user_id = await conversation_validator.validate(conversation_id, cookies_str)
+    user_id, github_user_id = await conversation_validator.validate(
+        conversation_id, cookies_str
+    )

    settings_store = await SettingsStoreImpl.get_instance(config, user_id)
    settings = await settings_store.load()
@@ -46,7 +52,7 @@ async def connect(connection_id: str, environ):
        )

    event_stream = await conversation_manager.join_conversation(
-        conversation_id, connection_id, settings, user_id
+        conversation_id, connection_id, settings, user_id, github_user_id
    )

    agent_state_changed = None
@@ -54,10 +60,7 @@ async def connect(connection_id: str, environ):
    async for event in async_stream:
        if isinstance(
            event,
-            (
-                NullAction,
-                NullObservation,
-            ),
+            (NullAction, NullObservation, RecallAction, RecallObservation),
        ):
            continue
        elif isinstance(event, AgentStateChangedObservation):
@@ -10,7 +10,12 @@ from openhands.events.action.message import MessageAction
 from openhands.integrations.github.github_service import GithubServiceImpl
 from openhands.integrations.provider import ProviderType
 from openhands.runtime import get_runtime_cls
-from openhands.server.auth import get_provider_tokens, get_access_token, get_github_user_id
+from openhands.server.auth import (
+    get_access_token,
+    get_github_user_id,
+    get_provider_tokens,
+    get_user_id,
+)
 from openhands.server.data_models.conversation_info import ConversationInfo
 from openhands.server.data_models.conversation_info_result_set import (
    ConversationInfoResultSet,
@@ -73,12 +78,12 @@ async def _create_new_conversation(
        logger.warn('Settings not present, not starting conversation')
        raise MissingSettingsError('Settings not found')

-    session_init_args['github_token'] = token or SecretStr('')
+    session_init_args['provider_token'] = token
    session_init_args['selected_repository'] = selected_repository
    session_init_args['selected_branch'] = selected_branch
    conversation_init_data = ConversationInitData(**session_init_args)
    logger.info('Loading conversation store')
-    conversation_store = await ConversationStoreImpl.get_instance(config, user_id)
+    conversation_store = await ConversationStoreImpl.get_instance(config, user_id, None)
    logger.info('Conversation store loaded')

    conversation_id = uuid.uuid4().hex
@@ -100,7 +105,8 @@ async def _create_new_conversation(
        ConversationMetadata(
            conversation_id=conversation_id,
            title=conversation_title,
-            github_user_id=user_id,
+            user_id=user_id,
+            github_user_id=None,
            selected_repository=selected_repository,
            selected_branch=selected_branch,
        )
@@ -122,7 +128,10 @@ async def _create_new_conversation(
            image_urls=image_urls or [],
        )
    await conversation_manager.maybe_start_agent_loop(
-        conversation_id, conversation_init_data, user_id, initial_message_action
+        conversation_id,
+        conversation_init_data,
+        user_id,
+        initial_user_msg=initial_message_action,
    )
    logger.info(f'Finished initializing conversation {conversation_id}')

@@ -158,7 +167,7 @@ async def new_conversation(request: Request, data: InitSessionRequest):
    try:
        # Create conversation with initial message
        conversation_id = await _create_new_conversation(
-            user_id,
+            get_user_id(request),
            github_token,
            selected_repository,
            selected_branch,
@@ -197,7 +206,7 @@ async def search_conversations(
    limit: int = 20,
 ) -> ConversationInfoResultSet:
    conversation_store = await ConversationStoreImpl.get_instance(
-        config, get_github_user_id(request)
+        config, get_user_id(request), get_github_user_id(request)
    )
    conversation_metadata_result_set = await conversation_store.search(page_id, limit)

@@ -216,7 +225,7 @@ async def search_conversations(
        conversation.conversation_id for conversation in filtered_results
    )
    running_conversations = await conversation_manager.get_running_agent_loops(
-        get_github_user_id(request), set(conversation_ids)
+        get_user_id(request), set(conversation_ids)
    )
    result = ConversationInfoResultSet(
        results=await wait_all(
@@ -236,7 +245,7 @@ async def get_conversation(
    conversation_id: str, request: Request
 ) -> ConversationInfo | None:
    conversation_store = await ConversationStoreImpl.get_instance(
-        config, get_github_user_id(request)
+        config, get_user_id(request), get_github_user_id(request)
    )
    try:
        metadata = await conversation_store.get_metadata(conversation_id)
@@ -252,7 +261,7 @@ async def update_conversation(
    request: Request, conversation_id: str, title: str = Body(embed=True)
 ) -> bool:
    conversation_store = await ConversationStoreImpl.get_instance(
-        config, get_github_user_id(request)
+        config, get_user_id(request), get_github_user_id(request)
    )
    metadata = await conversation_store.get_metadata(conversation_id)
    if not metadata:
@@ -268,7 +277,7 @@ async def delete_conversation(
    request: Request,
 ) -> bool:
    conversation_store = await ConversationStoreImpl.get_instance(
-        config, get_github_user_id(request)
+        config, get_user_id(request), get_github_user_id(request)
    )
    try:
        await conversation_store.get_metadata(conversation_id)
@@ -90,30 +90,38 @@ async def store_settings(
                    existing_settings.user_consents_to_analytics
                )

-            if existing_settings.secrets_store:
-                existing_providers = [
-                    provider.value
-                    for provider in existing_settings.secrets_store.provider_tokens
-                ]
-
-                # Merge incoming settings store with the existing one
-                for provider, token_value in settings.provider_tokens.items():
-                    if provider in existing_providers and not token_value:
-                        provider_type = ProviderType(provider)
-                        existing_token = (
-                            existing_settings.secrets_store.provider_tokens.get(
-                                provider_type
-                            )
-                        )
-                        if existing_token and existing_token.token:
-                            settings.provider_tokens[provider] = (
-                                existing_token.token.get_secret_value()
-                            )
-
-            # Merge provider tokens with existing ones
-            if settings.unset_github_token:  # Only merge if not unsetting tokens
+            if settings.unset_github_token:
                settings.secrets_store.provider_tokens = {}
                settings.provider_tokens = {}
+            else:  # Only merge if not unsetting tokens
+                if settings.provider_tokens:
+                    if existing_settings.secrets_store:
+                        existing_providers = [
+                            provider.value
+                            for provider in existing_settings.secrets_store.provider_tokens
+                        ]
+
+                        # Merge incoming settings store with the existing one
+                        for provider, token_value in settings.provider_tokens.items():
+                            if provider in existing_providers and not token_value:
+                                provider_type = ProviderType(provider)
+                                existing_token = (
+                                    existing_settings.secrets_store.provider_tokens.get(
+                                        provider_type
+                                    )
+                                )
+                                if existing_token and existing_token.token:
+                                    settings.provider_tokens[provider] = (
+                                        existing_token.token.get_secret_value()
+                                    )
+                else:  # nothing passed in means keep current settings
+                    provider_tokens = existing_settings.secrets_store.provider_tokens
+                    settings.provider_tokens = {
+                        provider.value: data.token.get_secret_value()
+                        if data.token
+                        else None
+                        for provider, data in provider_tokens.items()
+                    }

        # Update sandbox config with new settings
        if settings.remote_runtime_resource_factor is not None:
@@ -15,7 +15,8 @@ from openhands.core.schema.agent import AgentState
 from openhands.events.action import ChangeAgentStateAction, MessageAction
 from openhands.events.event import EventSource
 from openhands.events.stream import EventStream
-from openhands.microagent import BaseMicroAgent
+from openhands.memory.memory import Memory
+from openhands.microagent.microagent import BaseMicroAgent
 from openhands.runtime import get_runtime_cls
 from openhands.runtime.base import Runtime
 from openhands.runtime.impl.remote.remote_runtime import RemoteRuntime
@@ -52,7 +53,7 @@ class AgentSession:
        sid: str,
        file_store: FileStore,
        status_callback: Callable | None = None,
-        github_user_id: str | None = None,
+        user_id: str | None = None,
    ):
        """Initializes a new instance of the Session class

@@ -65,9 +66,9 @@ class AgentSession:
        self.event_stream = EventStream(sid, file_store)
        self.file_store = file_store
        self._status_callback = status_callback
-        self.github_user_id = github_user_id
+        self.user_id = user_id
        self.logger = OpenHandsLoggerAdapter(
-            extra={'session_id': sid, 'user_id': github_user_id}
+            extra={'session_id': sid, 'user_id': user_id}
        )

    async def start(
@@ -126,6 +127,15 @@ class AgentSession:
                agent_to_llm_config=agent_to_llm_config,
                agent_configs=agent_configs,
            )
+
+            repo_directory = None
+            if self.runtime and runtime_connected and selected_repository:
+                repo_directory = selected_repository.split('/')[-1]
+            self.memory = await self._create_memory(
+                selected_repository=selected_repository,
+                repo_directory=repo_directory,
+            )
+
            if github_token:
                self.event_stream.set_secrets(
                    {
@@ -231,7 +241,7 @@ class AgentSession:

        kwargs = {}
        if runtime_cls == RemoteRuntime:
-            kwargs['github_user_id'] = self.github_user_id
+            kwargs['user_id'] = self.user_id

        self.runtime = runtime_cls(
            config=config,
@@ -260,26 +270,14 @@ class AgentSession:
                )
            return False

-        repo_directory = None
        if selected_repository:
-            repo_directory = await call_sync_from_async(
+            await call_sync_from_async(
                self.runtime.clone_repo,
                github_token,
                selected_repository,
                selected_branch,
            )

-        if agent.prompt_manager:
-            agent.prompt_manager.set_runtime_info(self.runtime)
-            microagents: list[BaseMicroAgent] = await call_sync_from_async(
-                self.runtime.get_microagents_from_selected_repo, selected_repository
-            )
-            agent.prompt_manager.load_microagents(microagents)
-            if selected_repository and repo_directory:
-                agent.prompt_manager.set_repository_info(
-                    selected_repository, repo_directory
-                )
-
        self.logger.debug(
            f'Runtime initialized with plugins: {[plugin.name for plugin in self.runtime.plugins]}'
        )
@@ -342,6 +340,29 @@ class AgentSession:

        return controller

+    async def _create_memory(
+        self, selected_repository: str | None, repo_directory: str | None
+    ) -> Memory:
+        memory = Memory(
+            event_stream=self.event_stream,
+            sid=self.sid,
+            status_callback=self._status_callback,
+        )
+
+        if self.runtime:
+            # sets available hosts and other runtime info
+            memory.set_runtime_info(self.runtime)
+
+            # loads microagents from repo/.openhands/microagents
+            microagents: list[BaseMicroAgent] = await call_sync_from_async(
+                self.runtime.get_microagents_from_selected_repo, selected_repository
+            )
+            memory.load_user_workspace_microagents(microagents)
+
+            if selected_repository and repo_directory:
+                memory.set_repository_info(selected_repository, repo_directory)
+        return memory
+
    def _maybe_restore_state(self) -> State | None:
        """Helper method to handle state restore logic."""
        restored_state = None
@@ -8,6 +8,6 @@ class ConversationInitData(Settings):
    Session initialization data for the web environment - a deep copy of the global config is made and then overridden with this data.
    """

-    github_token: SecretStr | None = Field(default=None)
+    provider_token: SecretStr | None = Field(default=None)
    selected_repository: str | None = Field(default=None)
    selected_branch: str | None = Field(default=None)
@@ -61,7 +61,7 @@ class Session:
            sid,
            file_store,
            status_callback=self.queue_status_message,
-            github_user_id=user_id,
+            user_id=user_id,
        )
        self.agent_session.event_stream.subscribe(
            EventStreamSubscriber.SERVER, self.on_event, self.sid
@@ -123,11 +123,11 @@ class Session:

        agent = Agent.get_cls(agent_cls)(llm, agent_config)

-        github_token = None
+        provider_token = None
        selected_repository = None
        selected_branch = None
        if isinstance(settings, ConversationInitData):
-            github_token = settings.github_token
+            provider_token = settings.provider_token
            selected_repository = settings.selected_repository
            selected_branch = settings.selected_branch

@@ -140,7 +140,7 @@ class Session:
                max_budget_per_task=self.config.max_budget_per_task,
                agent_to_llm_config=self.config.get_agent_to_llm_config_map(),
                agent_configs=self.config.get_agent_configs(),
-                github_token=github_token,
+                github_token=provider_token,
                selected_repository=selected_repository,
                selected_branch=selected_branch,
                initial_message=initial_message,
@@ -43,7 +43,7 @@ class Settings(BaseModel):
        if context and context.get('expose_secrets', False):
            return llm_api_key.get_secret_value()

-        return pydantic_encoder(llm_api_key)
+        return pydantic_encoder(llm_api_key) if llm_api_key else None

    @staticmethod
    def _convert_token_value(
@@ -12,25 +12,36 @@ from openhands.utils.async_utils import wait_all


 class ConversationStore(ABC):
-    """
-    Storage for conversation metadata. May or may not support multiple users depending on the environment
-    """
+    """Storage for conversation metadata. May or may not support multiple users depending on the environment."""

    @abstractmethod
    async def save_metadata(self, metadata: ConversationMetadata) -> None:
-        """Store conversation metadata"""
+        """Store conversation metadata."""

    @abstractmethod
    async def get_metadata(self, conversation_id: str) -> ConversationMetadata:
-        """Load conversation metadata"""
+        """Load conversation metadata."""
+
+    async def validate_metadata(
+        self, conversation_id: str, user_id: str, github_user_id: str
+    ) -> bool:
+        """Validate that conversation belongs to the current user."""
+        # TODO: remove github_user_id after transition to Keycloak is complete.
+        metadata = await self.get_metadata(conversation_id)
+        if (not metadata.user_id and not metadata.github_user_id) or (
+            metadata.user_id != user_id and metadata.github_user_id != github_user_id
+        ):
+            return False
+        else:
+            return True

    @abstractmethod
    async def delete_metadata(self, conversation_id: str) -> None:
-        """delete conversation metadata"""
+        """Delete conversation metadata."""

    @abstractmethod
    async def exists(self, conversation_id: str) -> bool:
-        """Check if conversation exists"""
+        """Check if conversation exists."""

    @abstractmethod
    async def search(
@@ -49,6 +60,6 @@ class ConversationStore(ABC):
    @classmethod
    @abstractmethod
    async def get_instance(
-        cls, config: AppConfig, user_id: str | None
+        cls, config: AppConfig, user_id: str | None, github_user_id: str | None
    ) -> ConversationStore:
        """Get a store for the user represented by the token given"""
@@ -7,7 +7,7 @@ class ConversationValidator:
    """Storage for conversation metadata. May or may not support multiple users depending on the environment."""

    async def validate(self, conversation_id: str, cookies_str: str):
-        return None
+        return None, None


 conversation_validator_cls = os.environ.get(
@@ -85,8 +85,8 @@ class FileConversationStore(ConversationStore):
            try:
                conversations.append(await self.get_metadata(conversation_id))
            except Exception:
-                logger.error(
-                    f'Error loading conversation: {conversation_id}',
+                logger.warning(
+                    f'Could not load conversation metadata: {conversation_id}',
                )
        conversations.sort(key=_sort_key, reverse=True)
        conversations = conversations[start:end]
@@ -101,7 +101,7 @@ class FileConversationStore(ConversationStore):

    @classmethod
    async def get_instance(
-        cls, config: AppConfig, user_id: str | None
+        cls, config: AppConfig, user_id: str | None, github_user_id: str | None
    ) -> FileConversationStore:
        file_store = get_file_store(config.file_store, config.file_store_path)
        return FileConversationStore(file_store)
@@ -5,6 +5,7 @@ from datetime import datetime, timezone
@dataclass
 class ConversationMetadata:
    conversation_id: str
+    user_id: str | None
    github_user_id: str | None
    selected_repository: str | None
    selected_branch: str | None = None
@@ -1,25 +1,18 @@
 import os
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from itertools import islice

 from jinja2 import Template

 from openhands.controller.state.state import State
-from openhands.core.logger import openhands_logger
 from openhands.core.message import Message, TextContent
-from openhands.microagent import (
-    BaseMicroAgent,
-    KnowledgeMicroAgent,
-    RepoMicroAgent,
-    load_microagents_from_dir,
-)
-from openhands.runtime.base import Runtime
+from openhands.events.observation.agent import MicroagentKnowledge


@dataclass
 class RuntimeInfo:
-    available_hosts: dict[str, int]
-    additional_agent_instructions: str
+    available_hosts: dict[str, int] = field(default_factory=dict)
+    additional_agent_instructions: str = ''


@dataclass
@@ -32,75 +25,23 @@ class RepositoryInfo:

 class PromptManager:
    """
-    Manages prompt templates and micro-agents for AI interactions.
+    Manages prompt templates and includes information from the user's workspace micro-agents and global micro-agents.

-    This class handles loading and rendering of system and user prompt templates,
-    as well as loading micro-agent specifications. It provides methods to access
-    rendered system and initial user messages for AI interactions.
+    This class is dedicated to loading and rendering prompts (system prompt, user prompt).

    Attributes:
-        prompt_dir (str): Directory containing prompt templates.
-        microagent_dir (str): Directory containing microagent specifications.
-        disabled_microagents (list[str] | None): List of microagents to disable. If None, all microagents are enabled.
+        prompt_dir: Directory containing prompt templates.
    """

    def __init__(
        self,
        prompt_dir: str,
-        microagent_dir: str | None = None,
-        disabled_microagents: list[str] | None = None,
    ):
-        self.disabled_microagents: list[str] = disabled_microagents or []
        self.prompt_dir: str = prompt_dir
-        self.repository_info: RepositoryInfo | None = None
        self.system_template: Template = self._load_template('system_prompt')
        self.user_template: Template = self._load_template('user_prompt')
        self.additional_info_template: Template = self._load_template('additional_info')
        self.microagent_info_template: Template = self._load_template('microagent_info')
-        self.runtime_info = RuntimeInfo(
-            available_hosts={}, additional_agent_instructions=''
-        )
-
-        self.knowledge_microagents: dict[str, KnowledgeMicroAgent] = {}
-        self.repo_microagents: dict[str, RepoMicroAgent] = {}
-
-        if microagent_dir:
-            # This loads micro-agents from the microagent_dir
-            # which is typically the OpenHands/microagents (i.e., the PUBLIC microagents)
-
-            # Only load KnowledgeMicroAgents
-            repo_microagents, knowledge_microagents, _ = load_microagents_from_dir(
-                microagent_dir
-            )
-            assert all(
-                isinstance(microagent, KnowledgeMicroAgent)
-                for microagent in knowledge_microagents.values()
-            )
-            for name, microagent in knowledge_microagents.items():
-                if name not in self.disabled_microagents:
-                    self.knowledge_microagents[name] = microagent
-            assert all(
-                isinstance(microagent, RepoMicroAgent)
-                for microagent in repo_microagents.values()
-            )
-            for name, microagent in repo_microagents.items():
-                if name not in self.disabled_microagents:
-                    self.repo_microagents[name] = microagent
-
-    def load_microagents(self, microagents: list[BaseMicroAgent]) -> None:
-        """Load microagents from a list of BaseMicroAgents.
-
-        This is typically used when loading microagents from inside a repo.
-        """
-        openhands_logger.info('Loading microagents: %s', [m.name for m in microagents])
-        # Only keep KnowledgeMicroAgents and RepoMicroAgents
-        for microagent in microagents:
-            if microagent.name in self.disabled_microagents:
-                continue
-            if isinstance(microagent, KnowledgeMicroAgent):
-                self.knowledge_microagents[microagent.name] = microagent
-            elif isinstance(microagent, RepoMicroAgent):
-                self.repo_microagents[microagent.name] = microagent

    def _load_template(self, template_name: str) -> Template:
        if self.prompt_dir is None:
@@ -114,27 +55,6 @@ class PromptManager:
    def get_system_message(self) -> str:
        return self.system_template.render().strip()

-    def set_runtime_info(self, runtime: Runtime) -> None:
-        self.runtime_info.available_hosts = runtime.web_hosts
-        self.runtime_info.additional_agent_instructions = (
-            runtime.additional_agent_instructions
-        )
-
-    def set_repository_info(
-        self,
-        repo_name: str,
-        repo_directory: str,
-    ) -> None:
-        """Sets information about the GitHub repository that has been cloned.
-
-        Args:
-            repo_name: The name of the GitHub repository (e.g. 'owner/repo')
-            repo_directory: The directory where the repository has been cloned
-        """
-        self.repository_info = RepositoryInfo(
-            repo_name=repo_name, repo_directory=repo_directory
-        )
-
    def get_example_user_message(self) -> str:
        """This is the initial user message provided to the agent
        before *actual* user instructions are provided.
@@ -148,45 +68,6 @@ class PromptManager:

        return self.user_template.render().strip()

-    def enhance_message(self, message: Message) -> None:
-        """Enhance the user message with additional context.
-
-        This method is used to enhance the user message with additional context
-        about the user's task. The additional context will convert the current
-        generic agent into a more specialized agent that is tailored to the user's task.
-        """
-        if not message.content:
-            return
-
-        # if there were other texts included, they were before the user message
-        # so the last TextContent is the user message
-        # content can be a list of TextContent or ImageContent
-        message_content = ''
-        for content in reversed(message.content):
-            if isinstance(content, TextContent):
-                message_content = content.text
-                break
-
-        if not message_content:
-            return
-
-        triggered_agents = []
-        for name, microagent in self.knowledge_microagents.items():
-            trigger = microagent.match_trigger(message_content)
-            if trigger:
-                openhands_logger.info(
-                    "Microagent '%s' triggered by keyword '%s'",
-                    name,
-                    trigger,
-                )
-                # Create a dictionary with the agent and trigger word
-                triggered_agents.append({'agent': microagent, 'trigger_word': trigger})
-
-        if triggered_agents:
-            formatted_text = self.build_microagent_info(triggered_agents)
-            # Insert the new content at the start of the TextContent list
-            message.content.insert(0, TextContent(text=formatted_text))
-
    def add_examples_to_initial_message(self, message: Message) -> None:
        """Add example_message to the first user message."""
        example_message = self.get_example_user_message() or None
@@ -195,44 +76,28 @@ class PromptManager:
        if example_message:
            message.content.insert(0, TextContent(text=example_message))

-    def add_info_to_initial_message(
+    def build_workspace_context(
        self,
-        message: Message,
-    ) -> None:
-        """Adds information about the repository and runtime to the initial user message.
-
-        Args:
-            message: The initial user message to add information to.
-        """
-        repo_instructions = ''
-        assert (
-            len(self.repo_microagents) <= 1
-        ), f'Expecting at most one repo microagent, but found {len(self.repo_microagents)}: {self.repo_microagents.keys()}'
-        for microagent in self.repo_microagents.values():
-            # We assume these are the repo instructions
-            if repo_instructions:
-                repo_instructions += '\n\n'
-            repo_instructions += microagent.content
-
-        additional_info = self.additional_info_template.render(
+        repository_info: RepositoryInfo | None,
+        runtime_info: RuntimeInfo | None,
+        repo_instructions: str = '',
+    ) -> str:
+        """Renders the additional info template with the stored repository/runtime info."""
+        return self.additional_info_template.render(
+            repository_info=repository_info,
            repository_instructions=repo_instructions,
-            repository_info=self.repository_info,
-            runtime_info=self.runtime_info,
+            runtime_info=runtime_info,
        ).strip()

-        # Insert the new content at the start of the TextContent list
-        if additional_info:
-            message.content.insert(0, TextContent(text=additional_info))
-
    def build_microagent_info(
        self,
-        triggered_agents: list[dict],
+        triggered_agents: list[MicroagentKnowledge],
    ) -> str:
        """Renders the microagent info template with the triggered agents.

        Args:
-            triggered_agents: A list of dictionaries, each containing an "agent"
-                            (KnowledgeMicroAgent) and a "trigger_word" (str).
+            triggered_agents: A list of MicroagentKnowledge objects containing information
+                              about triggered microagents.
        """
        return self.microagent_info_template.render(
            triggered_agents=triggered_agents
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
openhands	bc27eae841	Remove CodeBLEU dependency and use tree-sitter directly	2025-03-17 16:45:19 +00:00
openhands	f0e5c81272	Merge main branch	2025-03-17 16:43:36 +00:00
Engel Nyst	a4b836b5f9	Don't try to send the new events in the UI (#7277 )	2025-03-17 14:50:22 +01:00
Xingyao Wang	a4d632498c	SWE-Gym rollout stability fix & using a validated SWE-Gym set (#7182 ) Co-authored-by: Robert Brennan <accounts@rbren.io> Co-authored-by: openhands <openhands@all-hands.dev> Co-authored-by: Engel Nyst <enyst@users.noreply.github.com> Co-authored-by: Graham Neubig <neubig@gmail.com>	2025-03-17 21:15:01 +08:00
Engel Nyst	4f017081fc	Quick fix docs (#7299 ) Co-authored-by: openhands <openhands@all-hands.dev>	2025-03-17 05:50:05 +00:00
Engel Nyst	51fb1fae88	RecallObservations (#7292 )	2025-03-17 03:18:22 +01:00
Graham Neubig	106b230fea	Update Slack invitation links (#7296 ) Co-authored-by: openhands <openhands@all-hands.dev>	2025-03-17 02:06:48 +00:00
Xingyao Wang	9b262dd057	fix retry on ConnectionError & retry on remote runtime by default (#7294 )	2025-03-17 01:18:54 +00:00
chuckbutkus	8074b261d3	Move current user_id to github_user_id and create a new user_id field (#7231 ) Co-authored-by: openhands <openhands@all-hands.dev> Co-authored-by: Rohit Malhotra <rohitvinodmalhotra@gmail.com>	2025-03-16 16:32:27 -04:00
dependabot[bot]	999a59f938	chore(deps): bump the version-all group with 5 updates (#7253 ) Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: openhands <openhands@all-hands.dev>	2025-03-16 19:51:08 +00:00
chuckbutkus	fbba57d3b5	Fix saving of settings (#7282 )	2025-03-16 19:06:46 +00:00
Engel Nyst	3f6c8a2338	Fix visual browsing (#7278 ) Co-authored-by: openhands <openhands@all-hands.dev>	2025-03-16 16:50:25 +01:00
Engel Nyst	dd09d46ccb	Remove DelegatorAgent (fix #7280 ) Co-authored-by: openhands <openhands@all-hands.dev>	2025-03-16 16:49:28 +01:00
tofarr	8897b45eeb	Fix for too much reaction in logs (#7276 )	2025-03-16 08:21:30 -06:00
Ryan H. Tran	30109e8f20	Separate tool descriptions to support models with limited description length (#7258 )	2025-03-16 09:48:13 +01:00
Engel Nyst	cc45f5d9c3	Add RecallActions and observations for retrieval of prompt extensions (#6909 ) Co-authored-by: openhands <openhands@all-hands.dev> Co-authored-by: Calvin Smith <email@cjsmith.io>	2025-03-15 21:48:37 +01:00
tofarr	e34a771e66	Fix for issue where initial command fails (#7254 )	2025-03-14 14:49:57 -06:00
tofarr	ec763f8105	Fix for error where credits is accessed even when billing is disabled (#7250 )	2025-03-14 15:10:54 +00:00
kjain14	4984bf6ee7	Merge pull request #6 from All-Hands-AI/main Main of OpenHands	2025-02-17 14:45:13 -05:00
Kush Dave Jain	92ddc1b46c	Merge branch 'All-Hands-AI-main'	2025-02-16 18:58:14 +00:00
Kush Dave Jain	367c8a9f83	Merging	2025-02-16 18:57:47 +00:00
Kush Dave Jain	09335d67be	Merging	2025-02-16 17:14:17 +00:00
kjain14	eb36426d33	Adding so file for codebleu	2025-02-11 06:57:07 -05:00
openhands	ace9e6e724	Update TestGenEval README to include dependency installation	2025-02-10 15:45:04 +00:00
Graham Neubig	1290a2599d	Update lock	2025-02-10 10:43:49 -05:00
openhands	513dd9791d	Restore testgeneval poetry group	2025-02-10 15:38:08 +00:00
Graham Neubig	fd53378d06	Rename eval-infer	2025-02-10 10:21:07 -05:00
Graham Neubig	4471002c79	Remove unneeded input	2025-02-09 08:25:08 -05:00
Graham Neubig	326e75e829	Remove prompt truncation	2025-02-09 07:31:36 -05:00
Graham Neubig	d8ad8babf6	Merge branch 'main' into main	2025-02-08 13:11:27 -05:00
kjain14	8782e3ae65	Merge pull request #3 from All-Hands-AI/main Merging master	2025-02-06 14:54:49 -05:00
Kush Dave Jain	eef0ed3410	Final prompt for final experiments	2025-02-06 19:48:04 +00:00
Kush Dave Jain	e7a8daf3ec	Fixing code to handle ablations	2025-02-04 03:10:11 +00:00
Kush Dave Jain	64abd4a95e	Ablation outputs	2025-01-30 21:12:42 +00:00
Kush Dave Jain	c7d575b4e1	Removing duplicate script	2025-01-28 21:29:30 +00:00
Kush Dave Jain	f781bc8343	Fix prompting	2025-01-28 21:11:35 +00:00
Kush Dave Jain	9f9a65c787	More updates	2025-01-20 19:39:17 +00:00
Graham Neubig	3355baea4c	Merge branch 'main' of github.com:kjain14/OpenHands into kjain14-main	2025-01-20 07:47:17 -05:00
Kush Dave Jain	8848e60c6d	Only top level filtering	2025-01-17 19:33:19 +00:00
Kush Dave Jain	d1e84093cc	Update filtering	2025-01-17 18:19:52 +00:00
Kush Dave Jain	3f0f13d335	Update prompt	2025-01-10 18:20:16 +00:00
Kush Dave Jain	219a134bb0	Refine prompt	2025-01-10 18:06:20 +00:00
Kush Dave Jain	efb525a463	Refine postprocessing	2025-01-09 22:53:15 +00:00
Kush Dave Jain	1ded123116	Reset to normal time	2025-01-08 19:08:29 +00:00
Kush Dave Jain	31b6967a87	Any and all pass	2025-01-08 18:04:50 +00:00
Graham Neubig	90422e5bfd	Update lock file	2024-12-25 16:43:59 -05:00
Graham Neubig	b47da9e894	Merge branch 'main' of github.com:All-Hands-AI/OpenHands into kjain14-main	2024-12-25 16:40:51 -05:00
openhands	3401bd610d	Update TestGenEval README with comprehensive information	2024-12-25 21:23:53 +00:00
Kush Dave Jain	77a153e42f	Final update, now working on all projects	2024-12-16 19:54:58 +00:00
Kush Dave Jain	fb9bc87e35	testgeneval deps	2024-12-12 15:50:35 +00:00
Kush Dave Jain	b685c67263	reset	2024-12-12 15:50:12 +00:00
Kush Dave Jain	2cd64bc636	Merge branch 'main' of https://github.com/kjain14/OpenHands	2024-12-11 22:25:32 +00:00
Kush Dave Jain	7c81deb132	Update README	2024-12-11 22:25:17 +00:00
kjain14	b19f735808	Merge pull request #2 from All-Hands-AI/main merging openhands	2024-12-11 17:06:55 -05:00
Kush Dave Jain	3af6025303	+ mutation testing	2024-12-11 22:05:46 +00:00
Kush Dave Jain	585dba9917	TestGenEval MVP	2024-12-11 20:36:20 +00:00
kjain14	bd66d09a33	Merge pull request #1 from All-Hands-AI/main Merge OpenHands	2024-12-06 11:14:34 -05:00
Kush Dave Jain	791b7f9f60	Cleaning to not OOM	2024-12-05 16:12:37 +00:00
Kush Dave Jain	30197e616b	Add option for starting point	2024-12-04 20:33:15 +00:00
Kush Dave Jain	f7f25319e3	Fixing testing dependencies	2024-12-04 14:34:01 +00:00
Kush Dave Jain	75fba59588	Readability metrics	2024-11-29 20:45:42 +00:00
Kush Dave Jain	280baa24e2	Licensing	2024-11-29 20:32:46 +00:00
Kush Dave Jain	c6206f5cf2	Initial pass for TestGenEval	2024-11-29 18:29:08 +00:00
Kush Dave Jain	7a4729c034	initial TestGenEval code	2024-11-29 17:57:57 +00:00