Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
134 lines (101 loc) · 4.41 KB

CONTRIBUTING.md

File metadata and controls

134 lines (101 loc) · 4.41 KB

Contributing to AutoHarness

Thank you for your interest in contributing to AutoHarness! This document covers the development setup, coding standards, and contribution process.

Development Setup

# Clone the repository
git clone https://github.com/aiming-lab/AutoHarness.git
cd autoharness

# Create a virtual environment (recommended)
python -m venv .venv
source .venv/bin/activate  # Linux/macOS
# .venv\Scripts\activate   # Windows

# Install in editable mode with dev dependencies
pip install -e ".[dev]"

Python version: 3.10 or higher is required.

Running Tests

# Quick run
make test

# Verbose output
make test-verbose

# Run a specific test file
python -m pytest tests/test_risk.py -v

# Run with coverage
python -m pytest tests/ --cov=autoharness --cov-report=term-missing

Code Style

We use ruff for both linting and formatting.

# Lint
make lint

# Auto-format
make format

# Type checking
make typecheck

Key rules:

  • Line length: 100 characters
  • Target: Python 3.10+
  • Ruff rule sets: E, F, W, I, N, UP, B, SIM, RUF
  • Type annotations are required for all public APIs
  • Strict mypy is enabled (strict = true in pyproject.toml)

Architecture Overview

autoharness/
├── core/            # Governance engine, permission engine, risk classifier,
│                    #   audit engine, constitution system, hook system,
│                    #   anti-distillation, sentiment analysis, feature flags
├── agent_loop.py    # Main agent execution loop integrating all subsystems
├── context/         # Context engine with 5-layer compaction
├── prompt/          # Prompt architecture and section framework
├── tools/           # Tool registry, concurrent orchestration, output budgets
├── skills/          # Skill system with two-layer injection
├── agents/          # Agent orchestration (fork, swarm, background, worktree),
│                    #   intelligent model routing (FAST/STANDARD/PREMIUM)
├── session/         # Session persistence, resume, cost tracking
├── tasks/           # Task system with dependency graph
├── rules/           # Hook profiles, 4-axis risk scoring
├── compiler/        # System prompt compilation
├── observability/   # Metrics and tracing
├── integrations/    # Anthropic, OpenAI, LangChain wrappers
├── templates/       # Built-in constitution and profile templates
├── marketplace/     # Skill/tool marketplace
├── cli/             # CLI commands (run, tools, skills, session, context)
└── wrap.py          # High-level wrapper API

Areas for Contribution

We welcome contributions in any area, but these are particularly impactful:

  • Anti-distillation (core/anti_distillation.py) -- improve decoy tool strategies and detection heuristics
  • Sentiment analysis (core/sentiment.py) -- expand frustration detection patterns and add support for additional languages
  • Model routing (agents/model_router.py) -- add new routing tiers, improve task complexity estimation
  • Feature flags (core/feature_flags.py) -- add persistent flag storage backends, admin UI integration
  • Context management (context/) -- improve compaction quality, add new compaction strategies
  • Constitution templates -- new industry-specific templates (e.g., healthcare, education)
  • Integration wrappers -- support for additional LLM providers (e.g., Gemini, Mistral)
  • Test coverage -- we currently have 920 tests; help us expand edge case coverage

Pull Request Process

  1. Fork the repository and create a feature branch from main.
  2. Write tests for any new functionality or bug fixes.
  3. Ensure all checks pass before submitting: 
    make lint format typecheck test
  4. Open a PR against main with a clear description of:
    • What the change does
    • Why it is needed
    • How it was tested
  5. A maintainer will review your PR. Address any feedback, then it will be merged.

Commit Messages

Use clear, imperative-mood commit messages:

  • Add session resume briefing support
  • Fix risk scoring for nested tool calls
  • Update context compaction thresholds

Reporting Issues

Open an issue on GitHub with:

  • Steps to reproduce
  • Expected vs. actual behavior
  • Python version and OS
  • Relevant logs or tracebacks

License

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