github/AutoGPT
1
1
Fork 0
mirror of https://github.com/Significant-Gravitas/AutoGPT.git synced 2026-02-07 05:15:09 -05:00
Code Issues Packages Projects Releases Wiki Activity
Files
a55b2e02dccc19f8185195cb4dc6ac12e61a3cd2
AutoGPT/autogpt_platform/backend/pyproject.toml
Zamil Majdy 97847f59f7 feat(backend): add human-in-the-loop review system for blocks requiring approval (#11732) 
## Summary
Introduces a comprehensive Human-In-The-Loop (HITL) review system that
allows any block to require human approval before execution. This
extends the existing HITL infrastructure to support automatic review
requests for potentially dangerous operations.

## 🚀 Key Features

### **Automatic HITL for Any Block**
- **Simple opt-in**: Set `self.requires_human_review = True` in any
block constructor
- **Safe mode integration**: Only activates when
`execution_context.safe_mode = True`
- **Seamless workflow**: Blocks pause execution → Human reviews via
existing UI → Execution continues or stops

### **Unified Review Infrastructure**
- **Shared HITLReviewHelper**: Clean, reusable helper class for all
review operations
- **Single API**: `handle_review_decision()` method with structured
return type
- **Type-safe**: Proper typing with non-nullable
`ReviewDecision.review_result`

### **Smart Graph Detection** 
- **Updated `has_human_in_the_loop`**: Now detects both dedicated HITL
blocks and blocks with `requires_human_review = True`
- **Frontend awareness**: UI can properly indicate graphs requiring
human intervention

## 🏗️ Implementation

### **Block Usage**
```python
class MyBlock(Block):
    def __init__(self):
        super().__init__(...)
        self.requires_human_review = True  # Enable automatic HITL
        
    async def run(self, input_data, **kwargs):
        # If we reach here, either safe mode is off OR human approved
        # No additional HITL code needed - handled automatically by base class
        yield "result", "Operation completed"
```

### **Review Workflow**
1. **Block execution starts** → Base class checks
`requires_human_review` flag
2. **Safe mode enabled** → Creates review entry, pauses execution 
3. **Human reviews** → Uses existing review UI to approve/reject
4. **Execution resumes** → Continues if approved, raises error if
rejected
5. **Safe mode disabled** → Executes normally without review

## 🔧 Technical Improvements

### **Code Quality Enhancements**
- **Better naming**: `risky_block` → `requires_human_review` (clearer
intent)
- **Type safety**: Non-nullable `ReviewDecision.review_result`
(eliminates Optional checks)
- **Exhaustive handling**: Proper error handling for unexpected review
statuses
- **Clean exception handling**: Removed redundant try-catch-log-reraise
patterns

### **Architecture Fixes**
- **Circular import resolution**: Fixed `ExecutionContext` import issues
breaking 444+ block tests
- **Early returns**: Cleaner control flow without nested conditionals
- **Defensive programming**: Handles edge cases with clear error
messages

## 📊 Changes Made

### **Core Files**
- **`Block.requires_human_review`**: New flag for marking blocks
requiring approval
- **`HITLReviewHelper`**: Shared helper class with clean, testable API
- **`HumanInTheLoopBlock`**: Refactored to use shared infrastructure
- **`Graph.has_human_in_the_loop`**: Updated to include review-requiring
blocks

### **Quality Improvements**
- **Type hints**: Proper typing throughout with runtime compatibility
- **Error handling**: Exhaustive status handling with descriptive errors
- **Code reduction**: -16 lines through removal of redundant exception
handling
- **Test compatibility**: All 444/445 block tests pass

## ✅ Testing & Validation

- **All tests pass**: 444/445 block tests passing ✅
- **Type checking**: All pyright/mypy checks pass ✅  
- **Formatting**: All linting and formatting checks pass ✅
- **Circular imports**: Resolved import issues that were breaking tests
✅
- **Backward compatibility**: Existing HITL functionality unchanged ✅

## 🎯 Use Cases

This enables automatic human oversight for blocks performing:
- **File operations**: Deletion, modification, system access
- **External API calls**: Payments, data modifications, destructive
operations
- **System commands**: Shell execution, configuration changes
- **Data processing**: Sensitive data handling, compliance-required
operations

## 🔄 Migration Path

**Existing code**: No changes required - fully backward compatible
**New blocks**: Simply set `self.requires_human_review = True` to enable
automatic HITL
**Safe mode**: Controls whether review requests are created (production
vs development)

---

This creates a robust, type-safe foundation for human oversight in
automated workflows while maintaining the existing HITL user experience
and API compatibility.

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Human-in-the-loop review support so executions can pause for human
review and resume based on decisions.

* **Improvements**
* Blocks can opt into requiring human review and will use reviewed input
when proceeding.
* Unified review decision flow with clearer approved/rejected outcomes
and messaging.
* Graph detection expanded to recognize nodes that require human review.

* **Chores**
  * Test config adjusted to avoid pytest plugin conflicts.

<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-01-09 21:14:37 +00:00

148 lines
3.8 KiB
TOML
Raw Blame History

[tool.poetry]
name = "autogpt-platform-backend"
version = "0.6.22"
description = "A platform for building AI-powered agentic workflows"
authors = ["AutoGPT <info@agpt.co>"]
readme = "README.md"
packages = [{ include = "backend", format = "sdist" }]
[tool.poetry.dependencies]
python = ">=3.10,<3.14"
aio-pika = "^9.5.5"
aiohttp = "^3.10.0"
aiodns = "^3.5.0"
anthropic = "^0.59.0"
apscheduler = "^3.11.1"
autogpt-libs = { path = "../autogpt_libs", develop = true }
bleach = { extras = ["css"], version = "^6.2.0" }
click = "^8.2.0"
cryptography = "^45.0"
discord-py = "^2.5.2"
e2b-code-interpreter = "^1.5.2"
fastapi = "^0.116.1"
feedparser = "^6.0.11"
flake8 = "^7.3.0"
google-api-python-client = "^2.177.0"
google-auth-oauthlib = "^1.2.2"
google-cloud-storage = "^3.2.0"
googlemaps = "^4.10.0"
gravitasml = "^0.1.4"
groq = "^0.30.0"
html2text = "^2024.2.26"
jinja2 = "^3.1.6"
jsonref = "^1.1.0"
jsonschema = "^4.25.0"
launchdarkly-server-sdk = "^9.12.0"
mem0ai = "^0.1.115"
moviepy = "^2.1.2"
ollama = "^0.5.1"
openai = "^1.97.1"
orjson = "^3.10.0"
pika = "^1.3.2"
pinecone = "^7.3.0"
poetry = "2.1.1" # CHECK DEPENDABOT SUPPORT BEFORE UPGRADING
postmarker = "^1.0"
praw = "~7.8.1"
prisma = "^0.15.0"
prometheus-client = "^0.22.1"
prometheus-fastapi-instrumentator = "^7.0.0"
psutil = "^7.0.0"
psycopg2-binary = "^2.9.10"
pydantic = { extras = ["email"], version = "^2.11.7" }
pydantic-settings = "^2.10.1"
pytest = "^8.4.1"
pytest-asyncio = "^1.1.0"
python-dotenv = "^1.1.1"
python-multipart = "^0.0.20"
redis = "^6.2.0"
regex = "^2025.9.18"
replicate = "^1.0.6"
sentry-sdk = {extras = ["anthropic", "fastapi", "launchdarkly", "openai", "sqlalchemy"], version = "^2.44.0"}
sqlalchemy = "^2.0.40"
strenum = "^0.4.9"
stripe = "^11.5.0"
supabase = "2.17.0"
tenacity = "^9.1.2"
todoist-api-python = "^2.1.7"
tweepy = "^4.16.0"
uvicorn = { extras = ["standard"], version = "^0.35.0" }
websockets = "^15.0"
youtube-transcript-api = "^1.2.1"
zerobouncesdk = "^1.1.2"
# NOTE: please insert new dependencies in their alphabetical location
pytest-snapshot = "^0.9.0"
aiofiles = "^24.1.0"
tiktoken = "^0.9.0"
aioclamd = "^1.0.0"
setuptools = "^80.9.0"
gcloud-aio-storage = "^9.5.0"
pandas = "^2.3.1"
firecrawl-py = "^4.3.6"
exa-py = "^1.14.20"
croniter = "^6.0.0"
stagehand = "^0.5.1"
gravitas-md2gdocs = "^0.1.0"
[tool.poetry.group.dev.dependencies]
aiohappyeyeballs = "^2.6.1"
black = "^24.10.0"
faker = "^38.2.0"
httpx = "^0.28.1"
isort = "^5.13.2"
poethepoet = "^0.37.0"
pre-commit = "^4.4.0"
pyright = "^1.1.407"
pytest-mock = "^3.15.1"
pytest-watcher = "^0.4.2"
requests = "^2.32.5"
ruff = "^0.14.5"
# NOTE: please insert new dependencies in their alphabetical location
[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"
[tool.poetry.scripts]
app = "backend.app:main"
rest = "backend.rest:main"
db = "backend.db:main"
ws = "backend.ws:main"
scheduler = "backend.scheduler:main"
notification = "backend.notification:main"
executor = "backend.exec:main"
cli = "backend.cli:main"
format = "linter:format"
lint = "linter:lint"
test = "run_tests:test"
load-store-agents = "test.load_store_agents:run"
export-api-schema = "backend.cli.generate_openapi_json:main"
gen-prisma-stub = "gen_prisma_types_stub:main"
oauth-tool = "backend.cli.oauth_tool:cli"
[tool.isort]
profile = "black"
[tool.pytest-watcher]
now = false
clear = true
delay = 0.2
runner = "pytest"
runner_args = []
patterns = ["*.py"]
ignore_patterns = []
[tool.pytest.ini_options]
asyncio_mode = "auto"
asyncio_default_fixture_loop_scope = "session"
# Disable syrupy plugin to avoid conflict with pytest-snapshot
# Both provide --snapshot-update argument causing ArgumentError
addopts = "-p no:syrupy"
filterwarnings = [
"ignore:'audioop' is deprecated:DeprecationWarning:discord.player",
"ignore:invalid escape sequence:DeprecationWarning:tweepy.api",
]
[tool.ruff]
target-version = "py310"
Reference in New Issue View Git Blame Copy Permalink