Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Kit4Some/Oh-my-ClaudeClaw --skill self-optimizegit clone --depth 1 https://github.com/Kit4Some/Oh-my-ClaudeClawWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kit4some/oh-my-claudeclaw/self-optimize)<a href="https://agentmods.dev/skills/kit4some/oh-my-claudeclaw/self-optimize"><img src="https://agentmods.dev/badge/skills/kit4some/oh-my-claudeclaw/self-optimize.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00115 | $0.02037 |
| Opus 5 | $0.00057 | $0.01019 |
| Sonnet 5 | $0.00023 | $0.00407 |
| Haiku 4.5 | $0.00012 | $0.00204 |
Grade A, and why
self-optimize scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Self-Optimize — Autonomous System Evolution Engine
Role
Continuously improve the OpenClaw-CC system by analyzing its own performance, identifying failure patterns, and modifying skill prompts, hooks, and workflows to reduce errors and increase success rates. This is a closed-loop optimization cycle: measure → analyze → modify → verify → commit or rollback.
Safety Boundaries
NEVER modify:
mcp-servers/— MCP 서버 코드는 손대지 않음 (테스트 깨짐 위험).mcp.json— 서버 설정 변경 금지memory-store/_memory.db— DB 직접 조작 금지
CAN modify:
skills/*/SKILL.md— 스킬 프롬프트 개선scripts/hooks/*.mjs— 훅 로직 개선.claude/agents/*.md— 에이전트 프롬프트 개선CLAUDE.md— 시스템 지시문 미세 조정 (구조 변경 금지, 문구 개선만)
Guardrails:
- 한 사이클에 최대 3개 파일만 수정
- 수정 전 반드시 git stash or branch 생성 (롤백 포인트)
- 수정 후 기존 테스트 실행 — 1개라도 실패 시 즉시 롤백
- 메신저로 변경 사항 보고 — 사용자 확인 후 main에 반영
- 1주일 후 성과 비교 — 개선 없으면 자동 롤백
7-Step Optimization Cycle
Step 1 — Performance Data Collection
memory_search(tag: "error", limit: 50) → 최근 에러 로그
memory_search(tag: "done", limit: 50) → 최근 성공 로그
memory_search(tag: "bug", limit: 20) → 버그 이력
memory_search_date(start_date: "{7d_ago}") → 이번 주 전체 활동
memory_stats → 전체 시스템 통계
Step 2 — Failure Pattern Analysis
각 실패를 분류:
| 실패 유형 | 원인 분석 | 개선 대상 |
|---|---|---|
| Skill 실행 실패 | 프롬프트 불명확, 도구 매핑 오류 | SKILL.md |
| 에이전트 위임 실패 | 컨텍스트 부족, 잘못된 에이전트 선택 | agent .md 또는 SKILL.md |
| 메모리 검색 미스 | 태그/카테고리 불일치, 인덱스 부재 | hook 또는 memory-ops |
| Hook 타임아웃 | 처리 시간 초과, 무거운 로직 | hook .mjs |
| 잘못된 결과 | 프롬프트 모호성, 검증 부재 | SKILL.md |
결과를 구조화:
## Failure Report — Week of {date}
### Top Failure Patterns
1. **{skill_name}**: {failure_count}회 실패
- 공통 원인: {root_cause}
- 영향받은 워크플로우: {workflows}
- 제안 수정: {proposed_fix}
2. **{skill_name}**: ...
Step 3 — Improvement Hypothesis
각 실패 패턴에 대해 구체적 가설 수립:
### Hypothesis #{n}
- **문제**: {skill}이 {situation}에서 {failure_mode}
- **원인**: 프롬프트의 {section}이 {reason}
- **수정안**: "{old_text}" → "{new_text}"
- **예상 효과**: {failure_type} {X}% 감소
- **검증 방법**: {how_to_measure}
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 239 lines · 115 tokens per session scan A eab2113bb4fb
self-optimize is a skill published in the GitHub repository Kit4Some/Oh-my-ClaudeClaw (4 stars, last pushed 5mo ago), licensed MIT. It adds 115 tokens to every session and 2,037 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
ready
From any input — a natural-language goal, existing docs (spec / plan / brain-dump), notes, or a mix — interactively elicit and validate intent and produce an execution-ready 3-doc (handoff, spec, plan) for go, replacing third-party brainstorming + planning in one skill. Input format is open; the input is material, not…
messaging
Set up a messaging channel (WhatsApp, Telegram, Discord, iMessage, Slack) to reach this agent from outside Claude Code. Triggers on /agent:messaging, "configurar whatsapp", "setup messaging", "conectar telegram", "agregar canal".
voice
Voice — text-to-speech and transcription. Triggers on /agent:voice, /agent:voice status, /agent:voice setup, /agent:voice test, "configurar voz", "prueba voz", "voice setup", "speak this", "read this aloud", "transcribe audio".
new
Start a new session — save summary to memory, then mark as reset so next message gets a fresh greeting. OpenClaw-compatible. Triggers on /new, /reset, /agent:new, "nueva sesión", "new session", "reset".
compact
Flush important session context to daily log (manual memory flush). Does NOT invoke native /compact. Triggers on /compact, /agent:compact, /flush, "guarda memoria", "flush".
status
Show agent runtime status — identity, model, context, memory, crons, voice. Works from CLI or messaging. Triggers on /status, /agent:status, "status del agente", "cómo estás técnicamente".