A starting baseline for open-source projects that use AI.

Curated conventions across four pillars. Clone the whole template, adopt by profile, or pick the pieces that fit.

See the use cases Compare profiles

Four pillars

Human-AI collaboration

AGENTS.md as the source of truth, multi-tool pointer files for Claude Code, Codex, GitHub Copilot, and Cursor, and the operational .ai/ workspace.

Open-source operations

README, LICENSE, contribution rules, security policy, support guide, issue and PR templates, code owners, funding, citation.

Engineering operations

CI workflows, versioning policy, Git workflow, branch protection, dependency automation, security baseline, OpenSSF Scorecard.

Knowledge and documentation

The strict in-repo docs/ reader and this knowledge layer — long-form guides, comparisons, blog, research.

Three profiles

Open Source

Governance, contribution, security, CI. No AI files.

Human-AI Collaboration

AGENTS.md, multi-tool pointers, .ai/ workspace. Drops into an existing repository.

Combined

All four pillars. The working reference in this repository.

Four adoption paths

  1. New project, with AI from day one — Combined profile.
  2. AI into an existing repository — Human-AI Collaboration profile alongside what already exists.
  3. New open-source project, no AI — Open Source profile only.
  4. Professionalising an existing open-source project — pull in the parts that are missing.

Walk-throughs for each path live in Use cases.

Two-layer architecture

The repository is the operational layer — workflows, AI state, governance, templates. This site is the knowledge layer — explanations, comparisons, tutorials, research. The repository itself serves as the working reference: every file and workflow runs here, not just on paper.