Skip to main content

Agency-Code: AI Developer Agent

An intelligent AI development assistant powered by Claude with built-in safety architecture.

Forked from: VRSEN/Agency-Code Enhanced with production-ready safety guardrails, auto-termination, and webhook integration.

Key Features: Multi-agent system (Coder + Planner) • Timeout monitoring • Runaway detection • WebSearch/WebFetch intelligence • Webhook-ready hooks • 181+ tests

📖 For detailed documentation, architecture, and integration examples, see ASKME.md


⚡ Quick Install (Recommended)

Requirements: Python 3.13 (Python 3.14 is not yet supported)

Windows Installation

# Step 1: Install agency-swarm framework
python -m pip install git+https://github.com/VRSEN/agency-swarm.git@main

# Step 2: Install agency-code
python -m pip install agency-code

# Step 3: Create .env file with your API key
"ANTHROPIC_API_KEY=sk-ant-your_key_here" | Out-File -FilePath .env -Encoding ASCII

# Step 4: Run aria
aria

Mac/Linux Installation

# Step 1: Install agency-swarm framework
pip install git+https://github.com/VRSEN/agency-swarm.git@main

# Step 2: Install agency-code
pip install agency-code

# Step 3: Create .env file with your API key
echo "ANTHROPIC_API_KEY=sk-ant-your_key_here" > .env

# Step 4: Run aria
aria

Get your API key: https://console.anthropic.com/settings/keys


🔧 Troubleshooting

Error: "aria is not recognized" (Windows)

Problem: Python Scripts folder is not in PATH.

Solution 1 - Add to PATH (Recommended):

# Run in PowerShell (restart terminal after)
[Environment]::SetEnvironmentVariable("Path", $env:Path + ";C:\Users\YOUR_USERNAME\AppData\Local\Programs\Python\Python313\Scripts", "User")

Solution 2 - Run with python:

python -m agency

Error: "command not found: aria" (Mac/Linux)

Problem: pip install directory not in PATH.

Solution 1 - Add to PATH:

# Add to ~/.bashrc or ~/.zshrc
export PATH="$HOME/.local/bin:$PATH"

# Then reload shell
source ~/.bashrc  # or source ~/.zshrc

Solution 2 - Run with python:

python -m agency

Error: "Fatal error in launcher" or broken pip

Problem: pip is pointing to wrong Python installation.

Solution: Use python -m pip instead:

# Windows
python -m pip install agency-code

# Mac/Linux
python3 -m pip install agency-code

Error: "ValueError: '3.14' is not a valid PythonVersion"

Problem: Python 3.14 is too new - agency-swarm only supports Python 3.13.

Solution: Install Python 3.13 from https://www.python.org/downloads/

Then use:

# Windows - use Python 3.13 specifically
py -3.13 -m pip install git+https://github.com/VRSEN/agency-swarm.git@main
py -3.13 -m pip install agency-code
py -3.13 -m agency

# Mac/Linux - ensure python3.13 is installed
python3.13 -m pip install git+https://github.com/VRSEN/agency-swarm.git@main
python3.13 -m pip install agency-code
python3.13 -m agency

Error: "Missing Anthropic API Key"

Problem: No .env file or API key not set.

Solution: Create .env file in the directory where you run aria:

Windows:

"ANTHROPIC_API_KEY=sk-ant-your_actual_key" | Out-File -FilePath .env -Encoding ASCII

Mac/Linux:

echo "ANTHROPIC_API_KEY=sk-ant-your_actual_key" > .env

Alternative - Set as environment variable:

# Windows (temporary - current session only)
$env:ANTHROPIC_API_KEY="sk-ant-your_key"

# Mac/Linux (temporary - current session only)
export ANTHROPIC_API_KEY="sk-ant-your_key"

🚀 Manual Installation (Development)

Prerequisites

Installation

Mac/Linux

# 1. Clone the repository
git clone https://github.com/joeyjoe808/Agent-C.git
cd Agent-C

# 2. Create and activate virtual environment
python3.13 -m venv .venv
source .venv/bin/activate

# 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt

# 4. Fix Anthropic compatibility (important!)
pip install git+https://github.com/openai/openai-agents-python.git@main

Windows (PowerShell)

# 1. Clone the repository
git clone https://github.com/joeyjoe808/Agent-C.git
cd Agent-C

# 2. Create and activate virtual environment
python -m venv .venv
.venv\Scripts\activate.ps1

# 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt

# 4. Fix Anthropic compatibility (important!)
pip install git+https://github.com/openai/openai-agents-python.git@main

Windows (Command Prompt)

REM 1. Clone the repository
git clone https://github.com/joeyjoe808/Agent-C.git
cd Agent-C

REM 2. Create and activate virtual environment
python -m venv .venv
.venv\Scripts\activate.bat

REM 3. Install dependencies
pip install --upgrade pip
pip install -r requirements.txt

REM 4. Fix Anthropic compatibility (important!)
pip install git+https://github.com/openai/openai-agents-python.git@main

🔑 Configuration

Create .env File

The agent looks for a .env file in the root directory. Create it with your API key:

Option 1: Create from scratch

# Create .env file
echo "ANTHROPIC_API_KEY=your_api_key_here" > .env

# On Windows (PowerShell):
# "ANTHROPIC_API_KEY=your_api_key_here" | Out-File -FilePath .env -Encoding ASCII

Option 2: Use example template (if provided)

cp .env.example .env
nano .env  # or notepad .env on Windows

.env File Contents

# Required: Your Anthropic API key
ANTHROPIC_API_KEY=sk-ant-xxxxxxxxxxxxxxxxxxxxxxxxxxxxx

# Optional: Enable/disable safety session tracking (default: true)
USE_SAFE_SESSION=true

# Optional: Model selection (default: claude-haiku-4-5)
MODEL=anthropic/claude-haiku-4-5-20251001

Get your API key: https://console.anthropic.com/settings/keys


▶️ Run Agency-Code

If installed via pipx/pip (Quick Install):

# Run from anywhere
aria

If installed manually (Development):

# Activate virtual environment (if not already activated)
source .venv/bin/activate  # Mac/Linux
# .venv\Scripts\activate     # Windows

# Run the agent
python agency.py
# Or use: aria

# On Mac, use sudo if you get permission errors:
sudo python agency.py

You'll see:

[SafeSession] [OK] Session tracking enabled
[SafeSession] Session ID: abc123...

User: [Type your request here]

Example Requests

User: Create a FastAPI hello world app
User: Fix the bug in database.py line 42
User: Research best practices for JWT authentication
User: Help me refactor this function to be more readable

🛡️ Safety Features

Built-in Protection:

  • ✅ Timeout Monitoring: 30min session / 5min turn / 2min tool timeouts
  • ✅ Runaway Detection: Catches infinite loops (5+ same tool calls)
  • ✅ Auto-Termination: Optional auto-kill on timeout/runaway (disabled by default)
  • ✅ Graceful Cancellation: Press Ctrl+C to save state and exit cleanly

Webhook Integration:

  • ✅ Hook System: Triggers on every tool execution and agent handoff
  • ✅ Perfect for APIs: Use as backend for your web application
  • ✅ Real-time Tracking: Session metrics (tool calls, duration, reasoning steps)

See ASKME.md for webhook integration examples and detailed safety documentation.


🧪 Testing

# Run all tests
pytest tests/ -v

# Run safety architecture tests (44 tests)
pytest tests/test_session_metrics.py \
       tests/test_safe_session.py \
       tests/test_timeout_monitor.py \
       tests/test_runaway_detector.py \
       tests/test_background_monitor.py \
       tests/test_cancellation.py -v

📂 Project Structure

Agency-Code/
├── agency.py                 # Main entry point
├── .env                      # API keys (you create this)
├── requirements.txt          # Dependencies
│
├── agency_code_agent/        # Coder agent
├── planner_agent/            # Planner agent
├── safety/                   # Safety architecture
├── tools/                    # 14+ tool implementations
├── shared/                   # Utilities & hooks
└── tests/                    # 181+ automated tests

🔗 Links


📖 Documentation

For detailed information, see ASKME.md:

  • Complete feature documentation
  • Webhook integration examples
  • Safety architecture details
  • API integration patterns
  • Demo tasks and examples
  • Architecture diagrams

For developers, see AGENTS.md:

  • Repository structure and guidelines
  • Coding style and conventions
  • Testing guidelines
  • Commit and PR guidelines

🤝 Contributing

Fully open-source - build, refine, and improve as needed!

  1. Fork the repository
  2. Create a feature branch
  3. Write tests first (TDD)
  4. Run test suite
  5. Submit pull request

📝 License

MIT License - see LICENSE for details.


Questions? Open an issue or check ASKME.md for detailed docs!

Metadata

Release files for agency-code 1.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for agency-code 1.1.0
File Size Uploaded Provenance
agency_code-1.1.0.tar.gz 77.4 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for agency-code 1.1.0
File Interpreter ABI Platform
agency_code-1.1.0-py3-none-any.whl Python 3 none any Details

Total release size: 166.0 kB

Release files / agency_code-1.1.0.tar.gz

Download URL agency_code-1.1.0.tar.gz
Size 77.4 kB
Tags Source
SHA-256 checksum
How to use checksums
f2baf5221a259efe010a3c02317a306763c443f31e8451e38fad91d656f7aeda
BLAKE2b-256 checksum
How to use checksums
a5ea0dbb3a5d603fb0d5967fed5e971eeef61acc41c36513258fb20ace9c91f6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.0

Release files / agency_code-1.1.0-py3-none-any.whl

Download URL agency_code-1.1.0-py3-none-any.whl
Size 88.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
15512d09a2a5ec08b5a6024af949800ed6203f7fc739cbcfd90bee782dd5cd7d
BLAKE2b-256 checksum
How to use checksums
0a8b83c772126293c3bb42dc2dd341fcc6732eceec96cfbc73127cfcc49fa7b6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.0

Security for agency-code 1.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.1.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page