Merge commit '116ba199d1c0d35b87af59254d1249c4fdd1fde5' into improve-cli-colors

Co-authored-by: openhands <openhands@all-hands.dev>

dev_config/python/.pre-commit-config.yaml

@@ -40,7 +40,7 @@ repos:
     hooks:
       - id: mypy
         additional_dependencies:
-          [types-requests, types-setuptools, types-pyyaml, types-toml, types-docker, pydantic, lxml]
+          [types-requests, types-setuptools, types-pyyaml, types-toml, types-docker, pydantic, lxml, types-Markdown]
         # To see gaps add `--html-report mypy-report/`
         entry: mypy --config-file dev_config/python/mypy.ini openhands/
         always_run: true

frontend/package-lock.json

@@ -6072,6 +6072,60 @@
         "node": ">=14.0.0"
       }
     },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/@emnapi/core": {
+      "version": "1.4.3",
+      "inBundle": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "@emnapi/wasi-threads": "1.0.2",
+        "tslib": "^2.4.0"
+      }
+    },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/@emnapi/runtime": {
+      "version": "1.4.3",
+      "inBundle": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "tslib": "^2.4.0"
+      }
+    },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/@emnapi/wasi-threads": {
+      "version": "1.0.2",
+      "inBundle": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "tslib": "^2.4.0"
+      }
+    },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/@napi-rs/wasm-runtime": {
+      "version": "0.2.11",
+      "inBundle": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "@emnapi/core": "^1.4.3",
+        "@emnapi/runtime": "^1.4.3",
+        "@tybys/wasm-util": "^0.9.0"
+      }
+    },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/@tybys/wasm-util": {
+      "version": "0.9.0",
+      "inBundle": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "tslib": "^2.4.0"
+      }
+    },
+    "node_modules/@tailwindcss/oxide-wasm32-wasi/node_modules/tslib": {
+      "version": "2.8.0",
+      "inBundle": true,
+      "license": "0BSD",
+      "optional": true
+    },
     "node_modules/@tailwindcss/oxide-win32-arm64-msvc": {
       "version": "4.1.11",
       "resolved": "https://registry.npmjs.org/@tailwindcss/oxide-win32-arm64-msvc/-/oxide-win32-arm64-msvc-4.1.11.tgz",

frontend/src/context/conversation-subscriptions-provider.tsx

@@ -14,6 +14,7 @@ import {
   isStatusUpdate,
 } from "#/types/core/guards";
 import { AgentState } from "#/types/agent-state";
+import EventLogger from "#/utils/event-logger";
 import {
   renderConversationErroredToast,
   renderConversationCreatedToast,

openhands/agenthub/codeact_agent/codeact_agent.py

@@ -106,10 +106,15 @@ class CodeActAgent(Agent):
     def _get_tools(self) -> list['ChatCompletionToolParam']:
         # For these models, we use short tool descriptions ( < 1024 tokens)
         # to avoid hitting the OpenAI token limit for tool descriptions.
-        SHORT_TOOL_DESCRIPTION_LLM_SUBSTRS = ['gpt-', 'o3', 'o1', 'o4']
+        SHORT_TOOL_DESCRIPTION_LLM_SUBSTRS = ['gpt-4', 'o3', 'o1', 'o4']
 
         use_short_tool_desc = False
         if self.llm is not None:
+            # For historical reasons, previously OpenAI enforces max function description length of 1k characters
+            # https://community.openai.com/t/function-call-description-max-length/529902
+            # But it no longer seems to be an issue recently
+            # https://community.openai.com/t/was-the-character-limit-for-schema-descriptions-upgraded/1225975
+            # Tested on GPT-5 and longer description still works. But we still keep the logic to be safe for older models.
             use_short_tool_desc = any(
                 model_substr in self.llm.config.model
                 for model_substr in SHORT_TOOL_DESCRIPTION_LLM_SUBSTRS

openhands/agenthub/codeact_agent/tools/bash.py

@@ -1,3 +1,4 @@
+import re
 import sys
 
 from litellm import ChatCompletionToolParam, ChatCompletionToolParamFunctionChunk
@@ -37,7 +38,16 @@ _SHORT_BASH_DESCRIPTION = """Execute a bash command in the terminal.
 
 def refine_prompt(prompt: str):
     if sys.platform == 'win32':
-        return prompt.replace('bash', 'powershell')
+        # Replace 'bash' with 'powershell' including tool names like 'execute_bash'
+        # First replace 'execute_bash' with 'execute_powershell' to handle tool names
+        result = re.sub(
+            r'\bexecute_bash\b', 'execute_powershell', prompt, flags=re.IGNORECASE
+        )
+        # Then replace standalone 'bash' with 'powershell'
+        result = re.sub(
+            r'(?<!execute_)(?<!_)\bbash\b', 'powershell', result, flags=re.IGNORECASE
+        )
+        return result
     return prompt
 
 

openhands/cli/main.py

@@ -739,19 +739,3 @@ def run_cli_command(args):
         except Exception as e:
             print_formatted_text(f'Error during cleanup: {e}')
             sys.exit(1)
-
-
-def main():
-    """Main entry point for OpenHands CLI."""
-    from openhands.core.config import get_cli_parser
-
-    parser = get_cli_parser()
-    args = parser.parse_args()
-
-    if hasattr(args, 'version') and args.version:
-        import openhands
-
-        print(f'OpenHands CLI version: {openhands.get_version()}')
-        sys.exit(0)
-
-    run_cli_command(args)

openhands/cli/tui.py

@@ -6,6 +6,7 @@ import asyncio
 import contextlib
 import datetime
 import json
+import re
 import sys
 import threading
 import time
@@ -36,6 +37,7 @@ from openhands.events import EventSource, EventStream
 from openhands.events.action import (
     Action,
     ActionConfirmationStatus,
+    AgentFinishAction,
     ChangeAgentStateAction,
     CmdRunAction,
     MCPAction,
@@ -65,10 +67,16 @@ MAX_RECENT_THOUGHTS = 5
 # Color and styling constants
 COLOR_GOLD = '#FFD700'
 COLOR_GREY = '#808080'
+COLOR_SUCCESS_GREEN = '#00D787'  # Bright green for finish actions
+COLOR_AGENT_BLUE = '#5FAFFF'  # Soft blue for agent messages
+COLOR_FINISH_FRAME = '#00AF87'  # Darker green for finish action frames
 DEFAULT_STYLE = Style.from_dict(
     {
         'gold': COLOR_GOLD,
         'grey': COLOR_GREY,
+        'success-green': COLOR_SUCCESS_GREEN,
+        'agent-blue': COLOR_AGENT_BLUE,
+        'finish-frame': COLOR_FINISH_FRAME,
         'prompt': f'{COLOR_GOLD} bold',
     }
 )
@@ -252,7 +260,10 @@ def display_thought_if_new(thought: str) -> None:
 def display_event(event: Event, config: OpenHandsConfig) -> None:
     global streaming_output_text_area
     with print_lock:
-        if isinstance(event, CmdRunAction):
+        if isinstance(event, AgentFinishAction):
+            # Handle agent finish actions with special styling
+            display_agent_finish(event)
+        elif isinstance(event, CmdRunAction):
             # For CmdRunAction, display thought first, then command
             if hasattr(event, 'thought') and event.thought:
                 display_message(event.thought)
@@ -275,8 +286,8 @@ def display_event(event: Event, config: OpenHandsConfig) -> None:
 
         if isinstance(event, MessageAction):
             if event.source == EventSource.AGENT:
-                # Check if this message content is a duplicate thought
-                display_thought_if_new(event.content)
+                # Display agent messages with distinctive styling
+                display_agent_message(event.content)
         elif isinstance(event, CmdOutputObservation):
             display_command_output(event.content)
         elif isinstance(event, FileEditObservation):
@@ -291,6 +302,29 @@ def display_event(event: Event, config: OpenHandsConfig) -> None:
             display_error(event.content)
 
 
+def process_markdown_for_terminal(text: str) -> str:
+    """
+    Process markdown syntax for terminal display.
+    This function handles common markdown patterns like bold, italic, code blocks, etc.
+    """
+    if not text:
+        return text
+
+    # Process bold text (**text**)
+    text = re.sub(r'\*\*(.*?)\*\*', r'\1', text)
+
+    # Process italic text (*text*)
+    text = re.sub(r'\*(.*?)\*', r'\1', text)
+
+    # Process inline code (`code`)
+    text = re.sub(r'`(.*?)`', r'\1', text)
+
+    # Process code blocks
+    text = re.sub(r'```(?:\w+)?\n(.*?)\n```', r'\1', text, flags=re.DOTALL)
+
+    return text
+
+
 def display_message(message: str) -> None:
     message = message.strip()
 
@@ -298,6 +332,76 @@ def display_message(message: str) -> None:
         print_formatted_text(f'\n{message}')
 
 
+def display_agent_message(message: str) -> None:
+    """Display a message from the agent with distinctive styling and markdown rendering."""
+    message = message.strip()
+
+    if message:
+        # Process markdown in the message
+        try:
+            # Process markdown for terminal display
+            processed_message = process_markdown_for_terminal(message)
+        except Exception:
+            # If markdown processing fails, use the original message
+            processed_message = message
+
+        container = Frame(
+            TextArea(
+                text=processed_message,
+                read_only=True,
+                style=COLOR_AGENT_BLUE,
+                wrap_lines=True,
+            ),
+            title='Agent Message',
+            style=f'fg:{COLOR_AGENT_BLUE}',
+        )
+        print_formatted_text('')
+        print_container(container)
+
+
+def display_agent_finish(event: AgentFinishAction) -> None:
+    """Display an agent finish action with distinctive styling and markdown rendering."""
+    # Determine the message to display
+    if event.final_thought:
+        message = event.final_thought
+    elif event.thought:
+        message = event.thought
+    else:
+        message = "All done! What's next on the agenda?"
+
+    # Add task completion status if available
+    if event.task_completed:
+        status_map = {
+            'true': '✅ Task completed successfully',
+            'partial': '⚠️ Task partially completed',
+            'false': '❌ Task could not be completed',
+        }
+        status_text = status_map.get(event.task_completed.value, '')
+        if status_text:
+            message = f'{status_text}\n\n{message}'
+
+    # Process markdown in the message
+    try:
+        # Process markdown for terminal display
+        processed_message = process_markdown_for_terminal(message)
+    except Exception:
+        # If markdown processing fails, use the original message
+        processed_message = message
+
+    container = Frame(
+        TextArea(
+            text=processed_message,
+            read_only=True,
+            style=COLOR_SUCCESS_GREEN,
+            wrap_lines=True,
+        ),
+        title='🎯 Agent Finished',
+        style=f'fg:{COLOR_FINISH_FRAME}',
+    )
+    print_formatted_text('')
+    print_container(container)
+
+
 def display_error(error: str) -> None:
     error = error.strip()
 

openhands/core/main.py

@@ -234,7 +234,7 @@ async def run_controller(
             file_path = config.save_trajectory_path
         os.makedirs(os.path.dirname(file_path), exist_ok=True)
         histories = controller.get_trajectory(config.save_screenshots_in_trajectory)
-        with open(file_path, 'w') as f:  # noqa: ASYNC101
+        with open(file_path, 'w') as f:  # noqa
             json.dump(histories, f, indent=4)
 
     return state

openhands/llm/fn_call_converter.py

@@ -383,7 +383,7 @@ Do NOT assume the environment is the same as in the example above.
 """
     example = example.lstrip()
 
-    return example
+    return refine_prompt(example)
 
 
 IN_CONTEXT_LEARNING_EXAMPLE_PREFIX = get_example_for_tools

openhands/resolver/issue_resolver.py

@@ -571,7 +571,7 @@ class IssueResolver:
         # checkout the repo
         repo_dir = os.path.join(self.output_dir, 'repo')
         if not os.path.exists(repo_dir):
-            checkout_output = subprocess.check_output(  # noqa: ASYNC101
+            checkout_output = subprocess.check_output(  # noqa
                 [
                     'git',
                     'clone',
@@ -584,7 +584,7 @@ class IssueResolver:
 
         # get the commit id of current repo for reproducibility
         base_commit = (
-            subprocess.check_output(['git', 'rev-parse', 'HEAD'], cwd=repo_dir)  # noqa: ASYNC101
+            subprocess.check_output(['git', 'rev-parse', 'HEAD'], cwd=repo_dir)  # noqa
             .decode('utf-8')
             .strip()
         )
@@ -596,7 +596,7 @@ class IssueResolver:
                 repo_dir, '.openhands_instructions'
             )
             if os.path.exists(openhands_instructions_path):
-                with open(openhands_instructions_path, 'r') as f:  # noqa: ASYNC101
+                with open(openhands_instructions_path, 'r') as f:  # noqa
                     self.repo_instruction = f.read()
 
         # OUTPUT FILE
@@ -605,7 +605,7 @@ class IssueResolver:
 
         # Check if this issue was already processed
         if os.path.exists(output_file):
-            with open(output_file, 'r') as f:  # noqa: ASYNC101
+            with open(output_file, 'r') as f:  # noqa
                 for line in f:
                     data = ResolverOutput.model_validate_json(line)
                     if data.issue.number == self.issue_number:
@@ -614,7 +614,7 @@ class IssueResolver:
                         )
                         return
 
-        output_fp = open(output_file, 'a')  # noqa: ASYNC101
+        output_fp = open(output_file, 'a')  # noqa
 
         logger.info(
             f'Resolving issue {self.issue_number} with Agent {AGENT_CLASS}, model {model_name}, max iterations {self.max_iterations}.'
@@ -633,20 +633,20 @@ class IssueResolver:
 
                 # Fetch the branch first to ensure it exists locally
                 fetch_cmd = ['git', 'fetch', 'origin', branch_to_use]
-                subprocess.check_output(  # noqa: ASYNC101
+                subprocess.check_output(  # noqa
                     fetch_cmd,
                     cwd=repo_dir,
                 )
 
                 # Checkout the branch
                 checkout_cmd = ['git', 'checkout', branch_to_use]
-                subprocess.check_output(  # noqa: ASYNC101
+                subprocess.check_output(  # noqa
                     checkout_cmd,
                     cwd=repo_dir,
                 )
 
                 base_commit = (
-                    subprocess.check_output(['git', 'rev-parse', 'HEAD'], cwd=repo_dir)  # noqa: ASYNC101
+                    subprocess.check_output(['git', 'rev-parse', 'HEAD'], cwd=repo_dir)  # noqa
                     .decode('utf-8')
                     .strip()
                 )

openhands/runtime/plugins/jupyter/__init__.py

@@ -69,7 +69,7 @@ class JupyterPlugin(Plugin):
 
             # Using synchronous subprocess.Popen for Windows as asyncio.create_subprocess_shell
             # has limitations on Windows platforms
-            self.gateway_process = subprocess.Popen(  # type: ignore[ASYNC101] # noqa: ASYNC101
+            self.gateway_process = subprocess.Popen(  # type: ignore[ASYNC101] # noqa
                 jupyter_launch_command,
                 stdout=subprocess.PIPE,
                 stderr=subprocess.STDOUT,
@@ -82,19 +82,19 @@ class JupyterPlugin(Plugin):
             output = ''
             while should_continue():
                 if self.gateway_process.stdout is None:
-                    time.sleep(1)  # type: ignore[ASYNC101] # noqa: ASYNC101
+                    time.sleep(1)  # type: ignore[ASYNC101] # noqa
                     continue
 
                 line = self.gateway_process.stdout.readline()
                 if not line:
-                    time.sleep(1)  # type: ignore[ASYNC101] # noqa: ASYNC101
+                    time.sleep(1)  # type: ignore[ASYNC101] # noqa
                     continue
 
                 output += line
                 if 'at' in line:
                     break
 
-                time.sleep(1)  # type: ignore[ASYNC101] # noqa: ASYNC101
+                time.sleep(1)  # type: ignore[ASYNC101] # noqa
                 logger.debug('Waiting for jupyter kernel gateway to start...')
 
             logger.debug(

openhands/runtime/utils/files.py

@@ -86,7 +86,7 @@ async def read_file(
         )
 
     try:
-        with open(whole_path, 'r', encoding='utf-8') as file:  # noqa: ASYNC101
+        with open(whole_path, 'r', encoding='utf-8') as file:  # noqa
             lines = read_lines(file.readlines(), start, end)
     except FileNotFoundError:
         return ErrorObservation(f'File not found: {path}')
@@ -127,7 +127,7 @@ async def write_file(
             os.makedirs(os.path.dirname(whole_path))
         mode = 'w' if not os.path.exists(whole_path) else 'r+'
         try:
-            with open(whole_path, mode, encoding='utf-8') as file:  # noqa: ASYNC101
+            with open(whole_path, mode, encoding='utf-8') as file:  # noqa
                 if mode != 'w':
                     all_lines = file.readlines()
                     new_file = insert_lines(insert, all_lines, start, end)

openhands/utils/prompt.py

@@ -4,6 +4,7 @@ from itertools import islice
 
 from jinja2 import Template
 
+from openhands.agenthub.codeact_agent.tools.bash import refine_prompt
 from openhands.controller.state.state import State
 from openhands.core.message import Message, TextContent
 from openhands.events.observation.agent import MicroagentKnowledge
@@ -91,7 +92,8 @@ class PromptManager:
             return Template(file.read())
 
     def get_system_message(self) -> str:
-        return self.system_template.render().strip()
+        system_message = self.system_template.render().strip()
+        return refine_prompt(system_message)
 
     def get_example_user_message(self) -> str:
         """This is an initial user message that can be provided to the agent

poetry.lock

@@ -1,4 +1,4 @@
-# This file is automatically @generated by Poetry 2.1.3 and should not be changed by hand.
+# This file is automatically @generated by Poetry 2.1.2 and should not be changed by hand.
 
 [[package]]
 name = "aiofiles"
@@ -3770,6 +3770,22 @@ http2 = ["h2 (>=3,<5)"]
 socks = ["socksio (==1.*)"]
 zstd = ["zstandard (>=0.18.0)"]
 
+[[package]]
+name = "httpx-aiohttp"
+version = "0.1.8"
+description = "Aiohttp transport for HTTPX"
+optional = false
+python-versions = ">=3.8"
+groups = ["main"]
+files = [
+    {file = "httpx_aiohttp-0.1.8-py3-none-any.whl", hash = "sha256:b7bd958d1331f3759a38a0ba22ad29832cb63ca69498c17735228055bf78fa7e"},
+    {file = "httpx_aiohttp-0.1.8.tar.gz", hash = "sha256:756c5e74cdb568c3248ba63fe82bfe8bbe64b928728720f7eaac64b3cf46f308"},
+]
+
+[package.dependencies]
+aiohttp = ">=3.10.0,<4"
+httpx = ">=0.27.0"
+
 [[package]]
 name = "httpx-sse"
 version = "0.4.0"
@@ -5214,6 +5230,22 @@ files = [
 [package.dependencies]
 cobble = ">=0.1.3,<0.2"
 
+[[package]]
+name = "markdown"
+version = "3.8.2"
+description = "Python implementation of John Gruber's Markdown."
+optional = false
+python-versions = ">=3.9"
+groups = ["main"]
+files = [
+    {file = "markdown-3.8.2-py3-none-any.whl", hash = "sha256:5c83764dbd4e00bdd94d85a19b8d55ccca20fe35b2e678a1422b380324dd5f24"},
+    {file = "markdown-3.8.2.tar.gz", hash = "sha256:247b9a70dd12e27f67431ce62523e675b866d254f900c4fe75ce3dda62237c45"},
+]
+
+[package.extras]
+docs = ["mdx_gh_links (>=0.2)", "mkdocs (>=1.6)", "mkdocs-gen-files", "mkdocs-literate-nav", "mkdocs-nature (>=0.6)", "mkdocs-section-index", "mkdocstrings[python]"]
+testing = ["coverage", "pyyaml"]
+
 [[package]]
 name = "markdown-it-py"
 version = "3.0.0"
@@ -10433,6 +10465,18 @@ files = [
 ]
 markers = {main = "extra == \"third-party-runtimes\""}
 
+[[package]]
+name = "types-markdown"
+version = "3.8.0.20250708"
+description = "Typing stubs for Markdown"
+optional = false
+python-versions = ">=3.9"
+groups = ["main"]
+files = [
+    {file = "types_markdown-3.8.0.20250708-py3-none-any.whl", hash = "sha256:d1f634931b463adf7603c012724b7e9e5eff976eb517dc700ebece2d6189b1ce"},
+    {file = "types_markdown-3.8.0.20250708.tar.gz", hash = "sha256:28690251fe90757f5a99cd671c79502bc2de07aef2d35fe54117c3b1c799804a"},
+]
+
 [[package]]
 name = "types-python-dateutil"
 version = "2.9.0.20250516"
@@ -11753,4 +11797,4 @@ third-party-runtimes = ["daytona", "e2b", "modal", "runloop-api-client"]
 [metadata]
 lock-version = "2.1"
 python-versions = "^3.12,<3.14"
-content-hash = "4640c66849d6436eed73826154e2d8cf88b456a4d1b71efb9438531245845826"
+content-hash = "0a2be134709df49a9e5132fdf0ec887f2a8cb99be0ed244349be638cbb48364b"

pyproject.toml

@@ -20,6 +20,7 @@ packages = [
 ]
 include = [
   "openhands/integrations/vscode/openhands-vscode-0.0.1.vsix",
+  "microagents/**/*",
 ]
 build = "build_vscode.py" # Build VSCode extension during Poetry build
 
@@ -41,6 +42,9 @@ numpy = "*"
 json-repair = "*"
 browsergym-core = "0.13.3"                         # integrate browsergym-core as the browsing interface
 html2text = "*"
+markdown = "*"                                     # For markdown processing in CLI
+types-Markdown = "*"                               # Type stubs for markdown
+deprecated = "*"
 pexpect = "*"
 jinja2 = "^3.1.3"
 python-multipart = "*"
@@ -97,6 +101,7 @@ e2b = { version = ">=1.0.5,<1.8.0", optional = true }
 modal = { version = ">=0.66.26,<1.2.0", optional = true }
 runloop-api-client = { version = "0.50.0", optional = true }
 daytona = { version = "0.24.2", optional = true }
+httpx-aiohttp = "^0.1.8"
 
 [tool.poetry.extras]
 third_party_runtimes = [ "e2b", "modal", "runloop-api-client", "daytona" ]
@@ -163,7 +168,7 @@ joblib = "*"
 swebench = { git = "https://github.com/ryanhoangt/SWE-bench.git", rev = "fix-modal-patch-eval" }
 
 [tool.poetry.scripts]
-openhands = "openhands.cli.main:main"
+openhands = "openhands.cli.entry:main"
 
 [tool.poetry.group.testgeneval.dependencies]
 fuzzywuzzy = "^0.18.0"

tests/unit/test_cli_thought_order.py

@@ -145,8 +145,8 @@ class TestThoughtDisplayOrder:
         # Verify that final thought is displayed
         mock_display_message.assert_called_once_with('This is a final thought.')
 
-    @patch('openhands.cli.tui.display_message')
-    def test_message_action_from_agent(self, mock_display_message):
+    @patch('openhands.cli.tui.display_agent_message')
+    def test_message_action_from_agent(self, mock_display_agent_message):
         """Test that MessageAction from agent is displayed."""
         config = MagicMock(spec=OpenHandsConfig)
 
@@ -156,8 +156,8 @@ class TestThoughtDisplayOrder:
 
         display_event(message_action, config)
 
-        # Verify that message is displayed
-        mock_display_message.assert_called_once_with('Hello from agent')
+        # Verify that agent message is displayed
+        mock_display_agent_message.assert_called_once_with('Hello from agent')
 
     @patch('openhands.cli.tui.display_message')
     def test_message_action_from_user_not_displayed(self, mock_display_message):

tests/unit/test_cli_tui.py

@@ -6,6 +6,8 @@ from openhands.cli.tui import (
     CustomDiffLexer,
     UsageMetrics,
     UserCancelledError,
+    display_agent_finish,
+    display_agent_message,
     display_banner,
     display_command,
     display_event,
@@ -26,6 +28,7 @@ from openhands.events import EventSource
 from openhands.events.action import (
     Action,
     ActionConfirmationStatus,
+    AgentFinishAction,
     CmdRunAction,
     MCPAction,
     MessageAction,
@@ -107,15 +110,15 @@ class TestDisplayFunctions:
         assert 'What do you want to build?' in message_text
         assert 'Type /help for help' in message_text
 
-    @patch('openhands.cli.tui.display_message')
-    def test_display_event_message_action(self, mock_display_message):
+    @patch('openhands.cli.tui.display_agent_message')
+    def test_display_event_message_action(self, mock_display_agent_message):
         config = MagicMock(spec=OpenHandsConfig)
         message = MessageAction(content='Test message')
         message._source = EventSource.AGENT
 
         display_event(message, config)
 
-        mock_display_message.assert_called_once_with('Test message')
+        mock_display_agent_message.assert_called_once_with('Test message')
 
     @patch('openhands.cli.tui.display_command')
     def test_display_event_cmd_action(self, mock_display_command):
@@ -182,6 +185,15 @@ class TestDisplayFunctions:
 
         mock_display_message.assert_called_once_with('Thinking about this...')
 
+    @patch('openhands.cli.tui.display_agent_finish')
+    def test_display_event_agent_finish(self, mock_display_agent_finish):
+        config = MagicMock(spec=OpenHandsConfig)
+        finish_action = AgentFinishAction(final_thought='Task completed')
+
+        display_event(finish_action, config)
+
+        mock_display_agent_finish.assert_called_once_with(finish_action)
+
     @patch('openhands.cli.tui.display_mcp_action')
     def test_display_event_mcp_action(self, mock_display_mcp_action):
         config = MagicMock(spec=OpenHandsConfig)
@@ -256,6 +268,42 @@ class TestDisplayFunctions:
         args, kwargs = mock_print.call_args
         assert message in str(args[0])
 
+    @patch('openhands.cli.tui.print_container')
+    @patch('openhands.cli.tui.print_formatted_text')
+    def test_display_agent_message(self, mock_print_formatted, mock_print_container):
+        message = 'Agent message'
+        display_agent_message(message)
+
+        mock_print_formatted.assert_called_once()
+        mock_print_container.assert_called_once()
+
+    @patch('openhands.cli.tui.print_container')
+    @patch('openhands.cli.tui.print_formatted_text')
+    def test_display_agent_finish_with_thought(
+        self, mock_print_formatted, mock_print_container
+    ):
+        finish_action = AgentFinishAction(thought='Final thought')
+
+        display_agent_finish(finish_action)
+
+        mock_print_formatted.assert_called_once()
+        mock_print_container.assert_called_once()
+
+    @patch('openhands.cli.tui.print_container')
+    @patch('openhands.cli.tui.print_formatted_text')
+    def test_display_agent_finish_with_task_completed(
+        self, mock_print_formatted, mock_print_container
+    ):
+        from openhands.events.action.agent import AgentFinishTaskCompleted
+
+        finish_action = AgentFinishAction()
+        finish_action.task_completed = AgentFinishTaskCompleted.TRUE
+
+        display_agent_finish(finish_action)
+
+        mock_print_formatted.assert_called_once()
+        mock_print_container.assert_called_once()
+
     @patch('openhands.cli.tui.print_container')
     def test_display_command_awaiting_confirmation(self, mock_print_container):
         cmd_action = CmdRunAction(command='echo test')

tests/unit/test_windows_prompt_refinement.py

@@ -0,0 +1,179 @@
+import sys
+from unittest.mock import patch
+
+import pytest
+
+from openhands.agenthub.codeact_agent.codeact_agent import CodeActAgent
+from openhands.core.config import AgentConfig
+from openhands.llm.llm import LLM
+
+# Skip all tests in this module if not running on Windows
+pytestmark = pytest.mark.skipif(
+    sys.platform != 'win32', reason='Windows prompt refinement tests require Windows'
+)
+
+
+@pytest.fixture
+def mock_llm():
+    """Create a mock LLM for testing."""
+    llm = LLM(config={'model': 'gpt-4', 'api_key': 'test'})
+    return llm
+
+
+@pytest.fixture
+def agent_config():
+    """Create a basic agent config for testing."""
+    return AgentConfig()
+
+
+def test_codeact_agent_system_prompt_no_bash_on_windows(mock_llm, agent_config):
+    """Test that CodeActAgent's system prompt doesn't contain 'bash' on Windows."""
+    # Create a CodeActAgent instance
+    agent = CodeActAgent(llm=mock_llm, config=agent_config)
+
+    # Get the system prompt
+    system_prompt = agent.prompt_manager.get_system_message()
+
+    # Assert that 'bash' doesn't exist in the system prompt (case-insensitive)
+    assert 'bash' not in system_prompt.lower(), (
+        f"System prompt contains 'bash' on Windows platform. "
+        f"It should be replaced with 'powershell'. "
+        f'System prompt: {system_prompt}'
+    )
+
+    # Verify that 'powershell' exists instead (case-insensitive)
+    assert 'powershell' in system_prompt.lower(), (
+        f"System prompt should contain 'powershell' on Windows platform. "
+        f'System prompt: {system_prompt}'
+    )
+
+
+def test_codeact_agent_tool_descriptions_no_bash_on_windows(mock_llm, agent_config):
+    """Test that CodeActAgent's tool descriptions don't contain 'bash' on Windows."""
+    # Create a CodeActAgent instance
+    agent = CodeActAgent(llm=mock_llm, config=agent_config)
+
+    # Get the tools
+    tools = agent.tools
+
+    # Check each tool's description and parameters
+    for tool in tools:
+        if tool['type'] == 'function':
+            function_info = tool['function']
+
+            # Check function description
+            description = function_info.get('description', '')
+            assert 'bash' not in description.lower(), (
+                f"Tool '{function_info['name']}' description contains 'bash' on Windows. "
+                f'Description: {description}'
+            )
+
+            # Check parameter descriptions
+            parameters = function_info.get('parameters', {})
+            properties = parameters.get('properties', {})
+
+            for param_name, param_info in properties.items():
+                param_description = param_info.get('description', '')
+                assert 'bash' not in param_description.lower(), (
+                    f"Tool '{function_info['name']}' parameter '{param_name}' "
+                    f"description contains 'bash' on Windows. "
+                    f'Parameter description: {param_description}'
+                )
+
+
+def test_in_context_learning_example_no_bash_on_windows():
+    """Test that in-context learning examples don't contain 'bash' on Windows."""
+    from openhands.agenthub.codeact_agent.tools.bash import create_cmd_run_tool
+    from openhands.agenthub.codeact_agent.tools.finish import FinishTool
+    from openhands.agenthub.codeact_agent.tools.str_replace_editor import (
+        create_str_replace_editor_tool,
+    )
+    from openhands.llm.fn_call_converter import get_example_for_tools
+
+    # Create a sample set of tools
+    tools = [
+        create_cmd_run_tool(),
+        create_str_replace_editor_tool(),
+        FinishTool,
+    ]
+
+    # Get the in-context learning example
+    example = get_example_for_tools(tools)
+
+    # Assert that 'bash' doesn't exist in the example (case-insensitive)
+    assert 'bash' not in example.lower(), (
+        f"In-context learning example contains 'bash' on Windows platform. "
+        f"It should be replaced with 'powershell'. "
+        f'Example: {example}'
+    )
+
+    # Verify that 'powershell' exists instead (case-insensitive)
+    if example:  # Only check if example is not empty
+        assert 'powershell' in example.lower(), (
+            f"In-context learning example should contain 'powershell' on Windows platform. "
+            f'Example: {example}'
+        )
+
+
+def test_refine_prompt_function_works():
+    """Test that the refine_prompt function correctly replaces 'bash' with 'powershell'."""
+    from openhands.agenthub.codeact_agent.tools.bash import refine_prompt
+
+    # Test basic replacement
+    test_prompt = 'Execute a bash command to list files'
+    refined_prompt = refine_prompt(test_prompt)
+
+    assert 'bash' not in refined_prompt.lower()
+    assert 'powershell' in refined_prompt.lower()
+    assert refined_prompt == 'Execute a powershell command to list files'
+
+    # Test multiple occurrences
+    test_prompt = 'Use bash to run bash commands in the bash shell'
+    refined_prompt = refine_prompt(test_prompt)
+
+    assert 'bash' not in refined_prompt.lower()
+    assert (
+        refined_prompt
+        == 'Use powershell to run powershell commands in the powershell shell'
+    )
+
+    # Test case sensitivity
+    test_prompt = 'BASH and Bash and bash should all be replaced'
+    refined_prompt = refine_prompt(test_prompt)
+
+    assert 'bash' not in refined_prompt.lower()
+    assert (
+        refined_prompt
+        == 'powershell and powershell and powershell should all be replaced'
+    )
+
+    # Test execute_bash tool name replacement
+    test_prompt = 'Use the execute_bash tool to run commands'
+    refined_prompt = refine_prompt(test_prompt)
+
+    assert 'execute_bash' not in refined_prompt.lower()
+    assert 'execute_powershell' in refined_prompt.lower()
+    assert refined_prompt == 'Use the execute_powershell tool to run commands'
+
+    # Test that words containing 'bash' but not equal to 'bash' are preserved
+    test_prompt = 'The bashful person likes bash-like syntax'
+    refined_prompt = refine_prompt(test_prompt)
+
+    # 'bashful' should be preserved, 'bash-like' should become 'powershell-like'
+    assert 'bashful' in refined_prompt
+    assert 'powershell-like' in refined_prompt
+    assert refined_prompt == 'The bashful person likes powershell-like syntax'
+
+
+def test_refine_prompt_function_on_non_windows():
+    """Test that the refine_prompt function doesn't change anything on non-Windows platforms."""
+    from openhands.agenthub.codeact_agent.tools.bash import refine_prompt
+
+    # Mock sys.platform to simulate non-Windows
+    with patch('openhands.agenthub.codeact_agent.tools.bash.sys.platform', 'linux'):
+        test_prompt = 'Execute a bash command to list files'
+        refined_prompt = refine_prompt(test_prompt)
+
+        # On non-Windows, the prompt should remain unchanged
+        assert refined_prompt == test_prompt
+        assert 'bash' in refined_prompt.lower()