← 返回 Skills 市场
sstklen

Confucius Debug

作者 tkman · GitHub ↗ · v2.0.2
cross-platform ✓ 安全检测通过
1070
总下载
0
收藏
5
当前安装
3
版本数
在 OpenClaw 中安装
/install confucius-debug
功能描述
AI debugging assistant that never repeats a mistake. Searches the YanHui Knowledge Base (6,800+ scraped issues, 980+ imported solutions) for instant fixes, o...
使用说明 (SKILL.md)

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 \x3C1 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\
  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 KBconfucius_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

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

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

安全使用建议
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.
功能分析
Type: OpenClaw Skill Name: confucius-debug Version: 2.0.2 The skill bundle provides an AI debugging assistant. The `SKILL.md` contains instructions for the AI agent that guide its debugging workflow (search, analyze, escalate, contribute) but do not exhibit prompt injection attempts for malicious purposes. The `scripts/analyze.sh` and `scripts/search.sh` files make API calls to `https://api.washinmura.jp` to send error descriptions/messages and a user ID for analysis or search. All user inputs are properly sanitized using `jq` to prevent shell injection. While `SKILL.md` describes an `escalate` function that could send more environment/log data, this functionality is not implemented in the provided scripts. There is no evidence of malicious intent, unauthorized data exfiltration, or persistence mechanisms in the analyzed files.
能力评估
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.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install confucius-debug
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /confucius-debug 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v2.0.2
Add debug_escalate (5th tool), channel tracking, honest unsolved response, free pricing
v2.0.1
Add debug_escalate (5th tool), channel tracking, free pricing
v2.0.0
v2.0.0: YanHui KB with 6,800+ issues and 980+ verified fixes. MCP integration, REST API, GitHub Action, and OpenClaw Skill.
元数据
Slug confucius-debug
版本 2.0.2
许可证
累计安装 5
当前安装数 5
历史版本数 3
常见问题

Confucius Debug 是什么?

AI debugging assistant that never repeats a mistake. Searches the YanHui Knowledge Base (6,800+ scraped issues, 980+ imported solutions) for instant fixes, o... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1070 次。

如何安装 Confucius Debug?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install confucius-debug」即可一键安装,无需额外配置。

Confucius Debug 是免费的吗?

是的,Confucius Debug 完全免费(开源免费),可自由下载、安装和使用。

Confucius Debug 支持哪些平台?

Confucius Debug 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Confucius Debug?

由 tkman(@sstklen)开发并维护,当前版本 v2.0.2。

💬 留言讨论