First, thank you for helping build the future of Agentic Legal Tech! ⚖️
To maintain the high standards required for legal-grade software, we ask all contributors to follow these guidelines.
- Fork the repository and create your branch from
main. - Environment: Use the provided
.env.example. Never commit your actual.envfile. - Testing: Ensure all logic passes the
pytestsuite, specifically focusing on the Hallucination-Loop Detection logic.
- AWS Credentials: Use IAM Roles or local profiles. We utilize
git-secretsto scan for accidental credential commits. - LLM Settings: Ensure all Bedrock calls have
opt-outfor model training where available to protect contract privacy. - Mock Data Only: Only use the anonymized
.txtor.pdffiles in thesamples/folder for pull request demonstrations.
- Type Hinting: We use strict Python type hinting to ensure the Supervisor Agent routes data deterministically.
- Emojis: Use emojis sparingly in logs and UI components to maintain a professional "Enterprise" aesthetic (e.g., 🔍 for Research, 🛡️ for Risk).
- Documentation: If you add a new Specialist Agent, you must update the architecture diagram or the Component Breakdown in the README.
By contributing to Factor-AI, you agree that your contributions will be licensed under the project's MIT License.