I created OAT to help maintaining consistency of coding-agent rule across projects , orgs, and personal preferences

Posted on by


Coding agents are great.

But maintaining rules across projects and within an organization is not easy.

Org policies, repo conventions, and developer personal preferences end up scattered across Cursor/Claude Code configs and random AGENTS.md files… and slowly contradict each other. Agents become inconsistent (and sometimes unsafe) because nobody knows what’s canonical.

So I created Org Agentic Toolkit (OAT): an open-source CLI to manage agent rules across org → project → user, and compile them into deterministic, validated outputs.

  • Org baseline (“constitution”) stays authoritative
  • Projects inherit explicitly
  • Users can add preferences as an overlay (without overriding org rules)
  • Compile + validate so it’s diffable, reviewable, and maintainable

Repo: https://github.com/alain-sv/org-agentic-toolkit

PyPI: https://pypi.org/project/org-agentic-toolkit/

Contributions welcome.



Source link