Contributing to AgentFlow

March 31, 2026 · View on GitHub

Thank you for your interest in contributing to AgentFlow! This guide will help you get started.

High-Impact Contribution Areas

1. PM Tool Adapters (Highest Priority)

AgentFlow currently works with Asana via MCP. The biggest unlock for the community is adapter support for additional tools:

  • GitHub Projects (highest priority — makes AgentFlow free to use)
  • Linear — popular with dev teams
  • Jira — enterprise support
  • Notion — for Notion-native teams

See adapters/ for the adapter interface and the existing Asana adapter as reference.

2. Stage Prompt Improvements

The prompts in prompts/ drive agent behavior at each pipeline stage. Improvements here directly improve pipeline quality:

  • Better adversarial review prompts
  • More thorough test stage prompts
  • Smarter research trigger heuristics
  • Improved decomposition quality

3. Documentation

  • Tutorials for specific project types (Next.js, Python, Flutter, etc.)
  • Video walkthroughs
  • Translations
  • Case studies

4. Conventions & Gap Registry

  • New gap discoveries (submit as issues first)
  • Improved gap fixes
  • Additional convention patterns for different tech stacks

How to Contribute

Reporting Issues

  1. Check existing issues first
  2. Include: what you expected, what happened, relevant config
  3. For gap discoveries: describe the failure mode, when it occurs, and proposed fix

Submitting Changes

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/your-feature
  3. Make your changes
  4. Test with a real pipeline run if possible
  5. Submit a PR with:
    • Clear description of the change
    • Which gap(s) it addresses (if applicable)
    • How you tested it

Writing Adapters

An adapter must implement these operations:

create_project(name, sections) → project_id
create_task(project_id, section, description) → task_id
move_task(task_id, section)
add_comment(task_id, body)
get_comments(task_id) → comments[]
search_tasks(query) → tasks[]
update_task_description(task_id, description)
get_sections(project_id) → sections[]
mark_complete(task_id)

Place your adapter in adapters/<tool-name>/ with:

  • README.md — setup instructions
  • Implementation files
  • Example configuration

Writing Prompts

Prompts follow a consistent structure:

  • Role description (who the agent is)
  • Input (what information they receive)
  • Process (step-by-step instructions)
  • Output (structured comments with machine-readable tags)
  • Rules (hard constraints)

All tags must be listed in conventions.md.

Code of Conduct

Be respectful, constructive, and focused on making AgentFlow better for everyone.

License

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