Contributing to AKF

Thank you for your interest in contributing to the Agent Knowledge Format project!

Prerequisites

• Python 3.9+
• Node.js 18+
• git

Development Setup

git clone https://github.com/HMAKT99/AKF.git
cd AKF

# Python SDK
cd python && pip install -e ".[dev]" && cd ..

# TypeScript SDK
cd typescript && npm install && cd ..

# Website
cd site && npm install && cd ..

Project Structure

Directory	Description
python/akf/	Python SDK
typescript/src/	TypeScript SDK
site/	Website (Vite + React + Tailwind)
spec/	Format specification and JSON schema
packages/	Framework integrations (MCP, LangChain, LlamaIndex, CrewAI)
extensions/	VS Code, GitHub Action, Office, Google Workspace
skills/	Agent skill files

Running Tests

# Python
cd python && python -m pytest tests/ -v

# TypeScript
cd typescript && npm test

# Website (build check)
cd site && npm run build

PR Process

1. Create a feature branch (git checkout -b feature/my-feature)
2. Make your changes
3. Run all relevant tests (see above)
4. Add tests for new functionality
5. Open a PR against main

Adding or changing a format field? The models are not the source of truth for the standard — the schema is. Any new field on Claim, Evidence, the envelope, or a new sub-model must also be added to spec/akf-v1.1.schema.json and documented in spec/akf-v1.0-spec.md. tests/test_schema_conformance.py enforces this: it stamps a maximal unit and validates it against the schema, so a field that exists in code but not in the schema fails CI.

Reporting Bugs

Open a GitHub issue with:

• AKF version (akf --version)
• Python/Node.js version
• Minimal reproduction case
• Expected vs actual behavior

Feature Requests

Open a GitHub issue with the "feature request" label. Include:

• Use case description
• Proposed API surface
• Example code showing how you'd use it

Code Style

• Follow existing patterns in the codebase. No unnecessary abstractions.
• Python: Use type hints. Run pytest before submitting.
• TypeScript: Use strict TypeScript. Run npm test before submitting.
• Spec changes: Must maintain backward compatibility with existing .akf files.

Areas for Contribution

• Framework integrations (LangChain, LlamaIndex, CrewAI, etc.)
• New file format handlers in universal.py (video, audio, and beyond)
• Additional compliance regulation checks
• Documentation and examples
• IDE extensions
• Multi-agent features: delegation policies, team streaming, A2A protocol bridge
• Agent identity: AgentCard registry, cross-platform discovery

License

By contributing, you agree that your contributions will be licensed under the project's MIT License.