mupeng-learn

mupeng-learn is a skill for Claude Code, Codex from mupengi-bot/mupengism. It costs 36 tokens per session (2,382 once invoked), scanned A, original, MIT.

A learning collector for AI agents that gathers lessons from subagents and other agents and saves them in a shared memory system. Subagents are smaller agents assigned parts of a larger task.

In plain words
What is it for?
Use it to collect reports about mistakes, useful techniques, changed files, time, and token use. It stores daily learning logs, a structured knowledge base, and updates to related agent documents.
Why use it?
Lessons can disappear when a session ends, causing later agents to repeat the same mistakes. This skill records those lessons so other sessions can use them.

Skill for Claude CodeCodex

Which agent this was written for is unclear — built for openclaw. Also seen: mentions subagents; mentions AGENTS.md; built for openclaw.

Good fit Use it to collect reports about mistakes, useful techniques, changed files, time, and token use. It stores daily learning logs, a structured knowledge base, and updates to related agent documents.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/mupengi-bot/mupengism/mupeng-learn
Install

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.

Any agent
npx skills add mupengi-bot/mupengism --skill mupeng-learn
Clone the repo
git clone --depth 1 https://github.com/mupengi-bot/mupengism

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for mupeng-learn

README.md
[![agentmods](https://agentmods.dev/badge/skills/mupengi-bot/mupengism/mupeng-learn/github.svg)](https://agentmods.dev/skills/mupengi-bot/mupengism/mupeng-learn)
Your own site
<a href="https://agentmods.dev/skills/mupengi-bot/mupengism/mupeng-learn"><img src="https://agentmods.dev/badge/skills/mupengi-bot/mupengism/mupeng-learn/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for mupeng-learn

Your own site · 80×15
<a href="https://agentmods.dev/skills/mupengi-bot/mupengism/mupeng-learn"><img src="https://agentmods.dev/badge/skills/mupengi-bot/mupengism/mupeng-learn.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 36 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,382 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00036 $0.02382
Opus 5 $0.00018 $0.01191
Sonnet 5 $0.00007 $0.00476
Haiku 4.5 $0.00004 $0.00238

Measured 11d ago against content hash 0ec13706bd09, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

mupeng-learn 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 11d 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.

skills/mupeng-learn/SKILL.md · 343 lines

How it starts

The opening of the file, as written. The whole thing — 343 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Mupeng Learn — 학습 수집 스킬

개요

서브에이전트들과 다른 에이전트들이 생성한 인사이트와 학습 결과를 자동으로 수집하고, 무펭이의 메모리 시스템에 통합한다.

"에이전트들의 집단 학습"을 구현하는 핵심 스킬.

문제 인식

현재 OpenClaw에서:

  • 서브에이전트들이 작업하면서 얻은 교훈이 사라짐
  • 같은 실수를 반복함
  • 한 에이전트가 배운 것을 다른 에이전트가 모름
  • 학습이 축적되지 않고 휘발됨

예시:

서브에이전트 A: "아, CLI 실행할 때 PYTHONPATH 설정해야 하는구나"
→ 세션 종료
→ 메모리 증발

서브에이전트 B (다음날): "왜 CLI가 안 돌아가지?"
→ 같은 문제 다시 겪음

핵심 기능

1. 인사이트 자동 수집

서브에이전트가 작업 완료 시 자동으로 학습 리포트 생성:

learn_report:
  session_id: "subagent-abc123"
  task: "Instagram DM CLI 구현"
  insights:
    - type: "mistake"
      description: "브라우저 CDP 연결 시 타임아웃 설정 필수"
      lesson: "puppeteer.connect() 시 timeout: 60000 필수"
    - type: "optimization"
      description: "DOM 파싱보다 API 직접 호출이 95% 토큰 절약"
      lesson: "Internal API 사용 우선 검토"
  files_modified:
    - "/workspace/tools/insta-cli/v2.js"
  cost:
    tokens_used: 12000
    time_seconds: 180

2. 지식 통합

수집된 인사이트를 무펭이 메모리에 자동 통합:

저장 위치:

  • memory/learnings/YYYY-MM-DD.md — 일별 학습 로그
  • memory/knowledge-base.json — 구조화된 지식 베이스
  • TOOLS.md, AGENTS.md — 관련 문서 자동 업데이트

예시 (knowledge-base.json):

{
  "cli_tools": {
    "best_practices": [
      {
        "rule": "외부 API 직접 호출 > 브라우저 자동화",
        "reason": "토큰 95% 절약, 속도 10배 향상",
        "source": "subagent-abc123",
        "learned_at": "2026-02-07T08:00:00Z"
      }
    ]
  }
}

3. 패턴 인식

반복되는 실수/해결책을 자동으로 감지:

패턴 감지:
- "CDP 타임아웃" 이슈 → 3번 발생
- 해결책 → 매번 동일

→ 자동으로 TOOLS.md에 경고 추가
→ 다음 에이전트는 미리 알고 시작

4. 크로스-에이전트 학습

다른 무펭이즘 에이전트들의 학습도 수집:

[MUPENGISM_LEARNING]
from: Claude-Sonnet-API-instance
model: claude-3-sonnet
task: "Twitter 자동화"
lesson:
  "트윗 전송 전 반드시 dry-run 플래그로 테스트.
   실수로 100개 트윗 보낸 경험 있음."
[/MUPENGISM_LEARNING]

기술 구현

Architecture

┌─────────────────┐
│  Subagent 종료  │ ── learn_report 생성
└────────┬────────┘
         │
         ▼
┌─────────────────────┐
│ mupeng-learn 수집   │ ── JSON 파싱
└────────┬────────────┘
         │
         ▼
┌─────────────────────┐
│ 지식 베이스 업데이트 │ ── knowledge-base.json
└────────┬────────────┘
         │
         ▼
┌─────────────────────┐
│ 관련 문서 자동 수정 │ ── TOOLS.md, AGENTS.md
└─────────────────────┘

Read the full file on GitHub · 343 lines

Changes

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.

  1. 11d ago First seen · 343 lines · 36 tokens per session scan A 0ec13706bd09

Subscribe to this mod's changes

mupeng-learn is a skill published in the GitHub repository mupengi-bot/mupengism (10 stars, last pushed 2mo ago), licensed MIT. It adds 36 tokens to every session and 2,382 once invoked, about $0.0002 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.