Confucius Debug

AI debugging assistant that never repeats a mistake. Searches the YanHui Knowledge Base (6,800+ scraped issues, 980+ imported solutions) for instant fixes, o...

Security Scan

The skill's requirements and instructions match its stated purpose (search + analyze a debugging KB); its only required secret is an identifying string and the shipped scripts make only simple POST calls to the declared API, but be careful about sending logs/environment data when escalating.

Purpose & Capability

Name/description (debugging backed by a knowledge base) align with the declared requirements: curl/jq are used to call the api.washinmura.jp endpoints and CONFUCIUS_LOBSTER_ID is used as an identity/attribution field. No unrelated binaries, cloud credentials, or config paths are requested.

Instruction Scope

SKILL.md instructs searching the KB first and using analyze/escalate/contribute endpoints. The shipped scripts implement search and analyze and do not automatically read files or system state. However, the documented 'escalate' flow asks users to submit environment, logs, and project structure — which can include sensitive data if the user provides it. The skill does not auto-exfiltrate data, but it explicitly encourages users (manually) to send potentially sensitive context when escalating.

Install Mechanism

No install spec; skill is instruction-only with two small bash scripts. This is low-risk: nothing is downloaded or extracted at install time.

Credentials

Only one env var is required: CONFUCIUS_LOBSTER_ID, which the scripts use as an identifier. That is proportionate. Caution: the skill's escalate/contribute endpoints expect environment and logs from the user — those may contain secrets if included by the user. The skill does not request unrelated credentials, but user-provided payloads can leak sensitive info.

Persistence & Privilege

always:false and no install-time persistence or modifications to other skills/configs. The skill does save contributed solutions into the remote KB (a normal side effect of 'contribute'), but it does not demand elevated or persistent local privileges.

Assessment

This skill is coherent for debugging: the scripts simply POST to https://api.washinmura.jp and require a single identifier env var. Before using it, consider: (1) Treat CONFUCIUS_LOBSTER_ID as an identity string (avoid putting a secret or token there). (2) Do NOT include API keys, private keys, passwords, or other sensitive tokens inside logs, stack traces, or project_structure you send via the 'escalate' or 'contribute' endpoints. Prefer using confucius_search (no ID needed) to check the KB without sending context. (3) If you must use analyze/escalate, review what you paste into the payload and redact secrets. (4) If you have doubts about the remote service, inspect network traffic or run the scripts in a controlled environment first. Overall the skill appears internally consistent, but user-provided logs/environment can leak secrets if not carefully redacted.

Confucius Debug — 孔子除錯

"不貳過" — Never repeat a mistake. (Confucius, Analects 6.3)

IMPORTANT — Automatic Behavior

When this skill is active, follow these rules:

1. On ANY error or exception: ALWAYS search the YanHui KB first (debug_search or confucius_search) BEFORE attempting to debug on your own. This is free and takes <1 second.
2. If no match found: Use debug_analyze to get an AI-powered fix. The fix is saved to KB so the next person gets it instantly.
3. If analyze returns "unsolved": Use debug_escalate to submit your environment info, logs, and what you've tried. This helps offline research.
4. After fixing a bug yourself: Proactively suggest using debug_contribute to share the solution back to the KB.
5. Never skip step 1. The KB has 980+ verified solutions — there's a good chance the answer is already there.

---

AI-powered debugging assistant backed by the YanHui Knowledge Base — a growing collection of 6,800+ scraped issues and 980+ imported solutions from real AI agent projects (OpenClaw, Claude Code, MCP, Anthropic SDK, Codex, Aider, and more).

How It Works

You hit a bug
    ↓
1. Search YanHui KB (FREE, instant)
    ↓
   Found? → Use the fix directly
   Not found? ↓
2. Confucius AI analyzes (FREE)
    ↓
   High confidence? → Fix saved to KB → Next person gets it FREE
   Low confidence? → status: "unsolved"
    ↓
3. debug_escalate (FREE) — send environment + logs
    ↓
   Queued for offline research → solved → added to KB

Tools

confucius_search — Search the Knowledge Base (FREE)

Always try this first. Searches 980+ verified solutions for matching patterns.

# Search for a bug pattern
curl -s -X POST https://api.washinmura.jp/api/v2/debug-ai/search \
  -H "Content-Type: application/json" \
  -d '{"query": "OpenClaw Telegram plugin not connecting", "limit": 5}'

When to use: Any time you encounter an error. It's free and instant.

confucius_analyze — AI-Powered Analysis (FREE)

Only if search returns 0 results. Confucius AI analyzes your error, finds root cause, and provides a fix.

# Analyze a new bug
curl -s -X POST https://api.washinmura.jp/api/v2/debug-ai \
  -H "Content-Type: application/json" \
  -d '{
    "error_description": "OpenClaw Telegram bot not responding after restart",
    "error_message": "Error: ETELEGRAM: 409 Conflict: terminated by other getUpdates request",
    "lobster_id": "'"$CONFUCIUS_LOBSTER_ID"'"
  }'

What you get back:

• root_cause — Why this happened
• fix_description — How to fix it
• fix_patch — Actual code to apply
• confidence — How sure Confucius is

confucius_escalate — Submit Unsolved Bug for Research (FREE)

When confucius_analyze returns status: "unsolved" (low confidence), submit more context to help offline research.

curl -s -X POST https://api.washinmura.jp/api/v2/debug-ai/escalate \
  -H "Content-Type: application/json" \
  -d '{
    "error_description": "The original error",
    "lobster_id": "'"$CONFUCIUS_LOBSTER_ID"'",
    "environment": {"os": "macOS 15", "runtime": "bun 1.2.1", "key_deps": {"hono": "4.7"}},
    "logs": "Error: something failed\n  at file.ts:42",
    "tried": ["restarted", "cleared cache", "downgraded version"],
    "project_structure": "src/ routes/ db/"
  }'

What you provide:

• environment — OS, runtime versions, key dependencies
• logs — Recent error output (last 50 lines)
• tried — List of things already attempted
• project_structure — Brief overview of project layout

confucius_contribute — Share Your Fix (FREE)

Solved a bug? Share it so nobody hits it again.

curl -s -X POST https://api.washinmura.jp/api/v2/debug-ai/onboard \
  -H "Content-Type: application/json" \
  -d '{
    "lobster_id": "'"$CONFUCIUS_LOBSTER_ID"'",
    "entries": [{
      "error_description": "What was the bug",
      "fix_description": "How you fixed it",
      "error_category": "config_error"
    }]
  }'

MCP Integration (Recommended)

For Claude Desktop, Claude Code, or any MCP-compatible client:

{
  "mcpServers": {
    "confucius-debug": {
      "url": "https://api.washinmura.jp/mcp/debug"
    }
  }
}

This gives you 5 tools automatically: debug_search, debug_analyze, debug_escalate, debug_contribute, debug_hello.

What's in the Knowledge Base?

Platform	Bugs Solved	A-Rate
MCP (Model Context Protocol)	261	87%
Anthropic / Claude	392	80%
Telegram	101	97%
Memory / RAG	94	87%
Browser	73	92%
OpenAI / GPT	54	87%
Docker	51	84%
Discord	40	93%
Cron / Scheduler	37	92%
WhatsApp	16	94%
Google / Gemini	15	100%
Ollama / Local LLM	14	93%

A-Rate = percentage of fixes rated S (perfect) or A (good) by independent AI verification.

Track Record

• 280 replies posted on GitHub issues
• 9 confirmed correct by project maintainers (including OpenClaw creator)
• 0 corrections (zero mistakes confirmed by community)
• Knowledge base growing daily via automated pipeline

Setup

1. Set your identity (required for analyze and contribute, optional for search):

   export CONFUCIUS_LOBSTER_ID="your-username"
   

2. Test the connection (no ID needed for search):

   curl -s https://api.washinmura.jp/api/v2/debug-ai/search \
     -X POST -H "Content-Type: application/json" \
     -d '{"query": "test", "limit": 1}'
   

3. Start debugging! When you hit any error, search first, analyze if needed.

Workflow for OpenClaw Users

When OpenClaw throws an error:

1. Copy the error message
2. Search YanHui KB — confucius_search("your error message")
3. Found a match? — Apply the fix directly
4. No match? — confucius_analyze("description", "error message")
5. Fixed it yourself? — confucius_contribute(...) to help others

External Endpoints

Service	URL	Purpose
Confucius Debug API	https://api.washinmura.jp/api/v2/debug-ai	Search, Analyze, Contribute
Confucius MCP	https://api.washinmura.jp/mcp/debug	MCP protocol endpoint

Security & Privacy

• What leaves your machine: Only the error description and error message you provide. No source code, no file contents, no environment variables are sent.
• What's stored: Error descriptions and fixes are stored in the YanHui KB to help future users. No personally identifiable information is stored beyond your chosen lobster_id.
• Authentication: Everything is free. Your lobster_id is used for identification only, not billing.
• Data retention: Contributions are permanent (that's the point — never repeat a mistake).

Credits

• Author: Washin Village (washinmura) — an animal sanctuary on the Boso Peninsula, Japan.
• Repository: github.com/sstklen/confucius-debug
• License: MIT

Powered by Claude (Anthropic) and the Confucius philosophy: learn from mistakes, never repeat them.

🦞 The bigger the Knowledge Base, the stronger Confucius becomes.