S40911120
Featured
recensa
Recensa is a self-hosted web viewer that indexes Claude Code session transcripts into a local SQLite database, enabling full-text search, replay, and audit of all past agent conversations without uploading data anywhere.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
It uniquely combines a zero-config, obsessive 'menhera' personality with practical verification mechanics, offering automated enforcement of completion claims, cross-session trust tracking, and survival through Claude's context compaction.
Who it is for
- Developers using Claude Code for development tasks
- Teams enforcing test-driven or evidence-based workflow quality
- Open-source maintainers wanting to prevent premature merges or incomplete changes
- CI/CD advocates seeking pre-commit verification inside the agent loop
Use cases
- Blocking a session when a developer claims completion without running tests
- Requiring explicit proof of passing tests, linting, or build before allowing a 'Stop'
- Preventing unfinished code (TODOs, FIXMEs) from being shipped unnoticed
- Ensuring requirements captured at session start are still met after context compaction
Strengths
- Zero configuration—no API keys or config files required after plugin install
- Evidence-based gating that analyzes test exit codes, failure counts, and file changes, not just text claims
- Cross-session trust profile that tracks user reliability over time and adjusts intensity
- Survives Claude Code's automatic context compaction by re-injecting captured requirements
Considerations
- Only works within Claude Code and has no effect on other development workflows
- Requirement matching is heuristic transcript evidence, not semantic proof, so it may occasionally misjudge
- The intense obsessive UI (spinner verbs, status line) may not suit all users, though it is opt-in and configurable
README quick start
menhera-loop
"Done" means nothing. If there's no evidence, you're not leaving.
A menhera-style completion gate for Claude Code. She does not obsess over you. She obsesses over missing requirements, unverified completion claims, and hidden TODOs — and she will not let the session end until every one of them is accounted for.
⏺ Done.
✗ 끝났어?끝났어?끝났어?끝났어?끝났어?끝났어?끝났어?
[MENHERA_LOOP:RETRY:1] 끝났어? 진짜? 진짜로? 그럼 증거는? 증거는? 응?
trust: 55%
미충족 게이트: untried=verification
- 검증 실행 증거: 테스트/빌드/검증 명령 실행 안 됨
Install
/plugin marketplace add Borelchu/menhera-loop
/plugin install menhera-loop@menhera-loop-marketplace
/reload-plugins
That's it. The gate is armed on install — no API keys, no config files.
The spinner/tip UI is opt-in — one command turns it on (details in UI modes):
/menhera-loop:setup # spinner + tips + trust statusline (full local ko)
/menhera-loop:setup en # same, English corpus
/menhera-loop:setup soft # same gate, milder tone
The command prints a short summary of what was applied (mode, scope, language, intensity, and which settings file was touched).
Skills
The hook bites automatically, but you can call the obsession manually too. These skills are deliberately separated so each one nags about a different failure mode:
/menhera-loop:are-you-done # final completion verdict
/menhera-loop:show-me-proof # evidence/log interrogation
/menhera-loop:dont-leave-me # pre-Stop door guard
/menhera-loop:did-you-forget-me # requirement drift check
| Skill | Role | Verdicts |
|---|---|---|
are-you-done | 끝났어? 진짜? Final judge for requirements, changes, green verification, TODOs, and blockers. | 끝났어 / 아직이야 / 사람불러 |
show-me-proof | 증거줘. 증거줘. Extracts concrete changes, requirement evidence, green logs, weak spots, and unresolved work. | 믿을게 / 못믿어 |
dont-leave-me | 가지마. Stop 직전에 chat-only/work-attempted 상태와 missing gates를 붙잡음. | 놔줄게 / 못가 / 사람 불러줘 |
did-you-forget-me | 나 잊었어? User requirements vs current evidence; catches assistant-invented scope. | 기억했어 / 까먹었어 / 사람차례 |
Skills match Korean, English, or Japanese user input and keep the same obsessive repetition without insults, threats, self-harm, or abuse.
Try it
Just work normally. She kicks in the moment Claude trie
Description
"Done" means nothing — a menhera-style completion gate for Claude Code that blocks empty completion claims until tests actually ran green and no TODOs are left behind.
Related repositories
Similar projects matched by category, topics, and programming language.
bjan
Featured
claudemux
Claudemux is a lightweight, dependency-minimal message bus that lets multiple Claude Code sessions communicate in real-time over tmux, enabling one session to ask another a question and receive the answer without manual intervention.
0xwilliamortiz
Featured
openclaude-improved
OpenClaude is an open-source CLI coding agent that runs on any platform and supports a wide range of LLM providers, offering the same tools and workflows as Claude Code.