Chainless

Contributing

Thank you for your interest in contributing to Chainless.
Every bug fix, improvement, documentation update, and new idea helps the project grow.

Chainless is a Python-only framework, so contributions revolve around Python code quality, tooling, FlowServer improvements, examples, and documentation.

How to Contribute

  1. Fork the repository and create a descriptive branch
    Example: feat/tool-tracking, fix/async-agent-bug, docs/flowserver.
  2. Implement your changes in a focused, well-scoped way.
    Avoid mixing unrelated edits in the same PR.
  3. Add or update tests when necessary.
  4. Run tests locally before opening a PR.
  5. Open a pull request against main with:
    • What you changed
    • Why it matters
    • Example usage if applicable

This keeps the review process fast and clean.

Project Structure (High-Level Overview)

Although Chainless evolves quickly, the core layout is usually:

  • chainless/agent — Agent system
  • chainless/tool — Tool wrapper
  • chainless/taskflow — TaskFlow engine
  • chainless/exp/server — FlowServer
  • chainless/schemas — Pydantic models
  • tests/ — test suite
  • apps/website/content/docs — documentation pages
  • examples/ — real-world usage examples

If you're unsure where your contribution fits, feel free to open an issue first.

Coding Guidelines

  • Python 3.9+ is required
  • Follow PEP8 standards
  • Prefer clear, readable code over clever tricks
  • Keep functions small and maintainable
  • Document public APIs and add docstrings
  • Avoid unnecessary abstractions — Chainless is meant to stay lightweight

Documentation Contributions

Docs live in the apps/website/content/docs/ directory as MDX files.

Improvements you can make:

  • Fix unclear explanations
  • Add missing examples
  • Expand FlowServer or TaskFlow tutorials
  • Clarify errors, edge cases, or tool behavior
  • Improve English phrasing or formatting

Documentation PRs are always welcome and often the easiest way to start contributing.

Tests

When fixing bugs or adding features:

  • Add a minimal test case that reproduces the issue
  • Place tests under tests/
  • Keep tests simple and descriptive
  • Use realistic input/output whenever possible

Tests help prevent regressions and ensure Chainless remains stable.

Issues & Feature Requests

You can open issues for:

  • Bugs
  • Missing features
  • Proposals for improvements
  • Questions about design decisions

Provide details like:

  • A reproducible code snippet
  • Expected vs actual behavior
  • Python version
  • Chainless version

Clear reports lead to faster fixes.

Go to GitHub Issues

Pull Requests

Before submitting a PR:

  • Code is formatted and readable
  • Tests added or updated
  • Docs updated if you changed public behavior
  • Commit messages are clear
  • PR description includes context and examples

Thank You

Whether you're improving documentation, fixing a bug, or shaping the future of the framework, your contribution is valuable.

If you want guidance on good starter tasks, feel free to open an issue — we’re happy to help you get started.

Web Analysis Workflow with Tools, Agents and Structured Output

A complete example showing how to build an autonomous website analysis workflow using Chainless. This example demonstrates tool integration, structured LLM outputs, TaskFlow orchestration and screenshot/HTML extraction using Playwright.