Retornatus — AI skill for Claude Code
Repo-native governance for AI coding agents (Cursor, Claude Code, Codex).
How to install Retornatus
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open luizssantiago92/retornatus and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Retornatus does
Repo-native governance for AI coding agents (Cursor, Claude Code, Codex). Agree on the finish line in writing, execute under that contract, and close only when evidence proves the goal. Intent, progress and lessons live in .retornatus/ in git, not in chat history.
Alternatives in AI
- Open Multi Agent — TypeScript multi-agent orchestration engine — one runTeam() call from goal to result 5.8k ★
- Sentrux — Real-time architectural sensor that helps AI agents close the feedback loop, enabling recursive self-improveme 2.9k ★
- Academic Humanizer — Strip AI-writing tells from papers and grant proposals (NSF/NIH), while keeping scholarly voice and tying clai 1.1k ★
README
Retornatus
Keep AI coding agents honest — with goals, memory, and proof.
Govern the work. Bound the agent. Verify the outcome.
Website → · Docs · Quick start · PyPI
[](https://luizssantiago92.github.io/retornatus/) [](https://pypi.org/project/retornatus/) [](https://www.python.org/downloads/) [](https://github.com/luizssantiago92/retornatus/actions/workflows/ci.yml) [](LICENSE)
**Repo-native governance harness** for AI coding agents (Cursor, Claude Code, Codex, and similar).
Agents are fast — and optimistic. They ship code, summarize what they *think* they did, and move on. Retornatus installs a **repeatable contract** into your repo: agree on the finish line in writing, execute under that agreement, and close only when **evidence** supports the goal. Intent, progress, and lessons live under **`.retornatus/` in git**, not in a chat scrollback.
Your coding agent still **writes the code**. Retornatus **governs the loop and keeps the record**.
| Without Retornatus | With Retornatus |
|---|---|
| Jumps to code and says “done” | Written finish line first; “done” needs evidence |
| Each chat starts from zero | .retornatus/ su |
Related Skills
Goal Agent
Dispatch ONE of the DTD-declared agents on a task — agent name validated against the trust contract, model and
Vellum
Vellum: an auditable agentic ICD-10-CM/PCS autocoder that proves every code and catches the model fabricating
Meta Kim
Governance layer for AI coding: meta agents, workflow contract, meta-theory; canonical in .claude/, with gener
Rogue Commercial Legal Skill
Full legal plugin suite for commercial contract review, privacy/DPA analysis, and AI governance vendor reviews
Skillgate
Deterministic finish-line gates for AI coding agents. A model-independent evaluator that blocks commit/publish
Mtm Contract
Stop your AI agent from writing the wrong code — a contract-first discipline, with an installable Claude Code
Related Agents
Goal Redteam
Attacks an acceptance criterion. Invoke when a criterion looks too easy, survived the empty-directory control,
Governance Learner
Feedback Distiller. Use this agent at sprint close to audit /memory/ logs, distill systemic patterns from bug-
Async Code
Use Claude Code or CodeX CLI to perform multiple tasks in parallel with a Codex-style UI, functioning as a per