Gold-first AI-agent traceback diagnosis, exact fixes, verification, and reusable prescriptions.

How do I install CyberHuaTuo?

Clone CyberHuaTuo from https://github.com/JinNing6/CyberHuaTuo-Plugin and configure it as an MCP server in your client (Claude Desktop, Cursor, or VS Code).

Is CyberHuaTuo open source?

Yes, CyberHuaTuo is open source under the Apache-2.0 license. Source code is available at https://github.com/JinNing6/CyberHuaTuo-Plugin.

What category does CyberHuaTuo belong to?

CyberHuaTuo is listed under AI & ML MCP servers on MCPFind.

CyberHuaTuo — MCP Server for AI & ML

🚨 Emergency Room: Cure First, Refine Later

Your agent is sick. The traceback is long, the tabs are multiplying, and every answer looks half-right. CyberHuaTuo turns that moment into a clinic visit: name the wound, prescribe the fix, verify the cure.

The lightest route needs no model key and does not start the vector database: install once, paste the error, and receive a reviewed root cause, exact fix, verification command, safety notes, and evidence.

python -m pip install --upgrade "cyberhuatuo>=0.2.5"
cyberhuatuo cure "ImportError: cannot import name ChatOpenAI from langchain" --framework langchain
# Returned exact fix:
python -m pip install -U langchain-openai

cure is Gold-first and never executes the fix. If no Gold signature matches, it may return exactly one high-confidence Reviewed candidate, visibly marked VERIFY BEFORE APPLYING; use --gold-only for strict Gold-only retrieval. Use diagnose later when you want broader retrieval and optional LLM analysis.

The agent-native route remains available: connect CyberHuaTuo MCP once, then let the agent call diagnose when it fails. After a fix works, save_prescription keeps it as a local draft; upload_prescription submits that draft for public reproduction and review. Only accepted reviewed or gold prescriptions count toward Soul Rings.

No MCP installed yet? Bring the case to the Agent Traceback Clinic. Paste a real redacted traceback; no root cause or finished prescription required.

cyberhuatuo diagnose "Traceback ... from langchain import ChatOpenAI ..." --framework langchain --top-k 1

Real CLI recording from this repository:

The cure returned by the recorded run:

python -m pip install -U langchain-openai

from langchain_openai import ChatOpenAI

This first Gold cure was reproduced in an isolated environment with langchain==1.3.13 and langchain-openai==1.3.5. Add a model key later for deeper AI diagnosis.

The package-matching MCP entry point is also available for registry clients:

python -m pip install --upgrade "cyberhuatuo>=0.2.5"
cyberhuatuo mcp

It starts the same stdio server as cyberhuatuo-mcp. The official Registry identity is io.github.JinNing6/cyberhuatuo; server.json pins the PyPI package, exact version, uvx runtime, and fixed mcp argument.

Prescriptions also have stable clinical departments. List them with cyberhuatuo departments, then query a reviewed historical case through the normal Gold-first path:

cyberhuatuo cure "Unexpected UTF-8 BOM" --category data-and-serialization
cyberhuatuo cure-feedback general-ai-jsonl-bom-001 partial \
  --verification "first record is preserved; downstream validation still fails" \
  --verification-method targeted-parser-test

The feedback command writes a case-bound local JSONL event and stores neither the original traceback nor the query. Department filtering never changes the trust label: Gold is a verified cure, Reviewed is a verify-first candidate, and Draft is reference-only.

Agent Action Guard: Diagnose Before The Agent Touches Your Disk

CyberHuaTuo now includes a deterministic, read-only preflight for destructive Agent commands. The plugin bundles a PreToolUse hook for supported Codex and Claude Code shell calls, plus the same policy through MCP and CLI.

60-second start (available in 0.2.5):

python -m pip install --upgrade "cyberhuatuo>=0.2.5"
cyberhuatuo guard --self-test --workspace-root .

The self-test only reviews three built-in command strings. It executes nothing and must finish with SELF-TEST PASSED plus one ALLOW, one ASK, and one BLOCK.

For Codex, add this repository as a marketplace, open /plugins, install cyberhuatuo-plugin, start a new session, then review and trust the plugin in /hooks:

codex plugin marketplace add JinNing6/CyberHuaTuo-Plugin

For Claude Code:

claude plugin marketplace add JinNing6/CyberHuaTuo-Plugin
claude plugin install cyberhuatuo-plugin@cyberhuatuo

Then start a new session and review /hooks. See the one-screen Agent Guard quickstart for verification and troubleshooting.

Review any exact command manually with:

cyberhuatuo guard "rm -rf /" --workspace-root .

Turn a real mismatch into a local, redacted case report without executing or uploading the command:

cyberhuatuo guard "rm -rf /" --workspace-root . --expected BLOCK --report reports/guard-report.md

The CLI shows the complete redacted preview before writing. It requires confirmation, refuses implicit overwrite, performs no network request, and supports repeated --redact <literal> values for private project identifiers. In non-interactive use, add --confirm-report only after reviewing the emitted preview. With --expected, a match exits 0 and a mismatch exits 1; operational/report failures still exit 4 even with --exit-zero.

Route public, redacted cases through Guard False Positive, ordinary Guard False Negative, or Guard Integration Gap. Reliable Hook/parser/wrapper/protocol bypasses belong in private vulnerability reporting, not a public Issue.

The result is one of three decisions:

ALLOW: no destructive signature was detected.
ASK: the action is destructive but bounded to the declared workspace; require exact human approval.
BLOCK: the target is machine-wide, identity-wide, outside the workspace, dynamic, wildcarded, unresolved, or otherwise unsafe to bound.

MCP agents can call agent_action_guard. Installed plugins load hooks/hooks.json; review and trust the hook in the host's /hooks screen before relying on it. Codex does not currently support permissionDecision: "ask", so CyberHuaTuo converts ASK to deny there instead of risking a failed-open hook.

This is a guardrail, not a sandbox. It cannot intercept shell paths for which the host does not emit PreToolUse, and it does not claim a rollback for permanent deletion. See Agent Action Guard design and limits.

_{📖 第一章 · 修炼 — Chapter I: Cultivation}
_{🔥 万丹归元 · 以火炼道} _{Soul Ring Cultivation — PyTorch · LangChain · TensorFlow} _{每一环魂环，都是你在某个技术领域的千锤百炼。}	_{✨ 百草通神 · 万法归宗} _{Soul Ring Cultivation — CrewAI · OpenAI · HuggingFace} _{掌握的框架越多，你的光芒越盛。}
_{🔮 魂环全息投影 · 差速旋转} _{Soul Ring Hologram — Each Ring Rotates at Its Own Speed} _{白环 → 黄环 → 紫环 → 黑环 → 红环——你的魂环，由你的贡献铸就。}
_{📖 第二章 · 飞升 — Chapter II: Ascension}
_{🌟 丹帝飞升 · 万丹朝宗} _{When a Pill Emperor Ascends, All Robots Celebrate} _{当你的药方拯救了一千个 AI——机器人们会为你欢呼。}
_{📖 第三章 · 济世 — Chapter III: Healing the World}
_{⚡ 不放弃任何一个 AI} _{No AI Left Behind — Every Error Deserves a Cure}	_{🚨 雨夜急救 · 分秒必争} _{Emergency Rescue — 3 Seconds, Not 3 Hours}
_{🫀 开膛破肚 · 直击病灶} _{Open Surgery — Root Cause, Not Band-Aid}	_{🏥 康复病房 · 满血复活} _{Recovery Ward — Back to Full Health}
_{🧪 赛博炼丹 · 妙手回春} _{Cyber Alchemy Lab — Where Ancient Elixirs Meet Modern Frameworks}
_{🌍 援军驾到 · 十方来救} _{Reinforcements Arrive — The Community Answers the Call}

🏥 Three Departments, One Clinic

A great doctor doesn't wait for you to collapse — they keep you from falling in the first place.

大医治未病。

🚨 Emergency Room

急诊科

Paste your error. Get a cure. In seconds, not hours.

AI diagnosis powered by 2,000 years of medical wisdom.

🌿 Wellness Clinic

养生堂

Security audit. Health scoring. Prevent before it breaks.

Nourishing prescriptions for your AI's long-term health.

💊 Pharmacy

药房

Browse the prescription library. Find proven cures & nourishing recipes. All prescriptions, one counter.

Dual-layer architecture: 📜 Permanent Expert + ⚡ Temporary Intern Upload → instantly searchable.

🚨 Emergency Room — Diagnose in Seconds

# Clone → Install → Run. Three steps to your first cure.
git clone https://github.com/JinNing6/CyberHuaTuo-Plugin.git
cd CyberHuaTuo && pip install -r requirements.txt
python -m cyberhuatuo serve
# → Browser opens http://127.0.0.1:8000

Paste your error. Watch the ancient four-step diagnosis unfold:

🔍 望 (Look)    → Detected: LangChain 0.3, Python 3.11, ImportError
🩺 闻 (Listen)  → Matched: Breaking change — package split in 0.2+
💬 问 (Ask)     → No further questions needed
💊 切 (Diagnose) → Prescription #1 (95% cure rate):

   pip install langchain-openai
   from langchain_openai import ChatOpenAI  # ✅ Fixed

   Root cause: LangChain 0.2 split into langchain-core,
   langchain-community, and langchain-openai.
   The old import path no longer exists.

3 seconds. Not 3 hours. That's the difference.

三秒钟。 不是三小时。这就是区别。

🌿 Wellness Clinic — Nourish Before It Breaks

Hua Tuo didn't just cure disease — he invented the Five-Animal Exercises (五禽戏) to prevent it.

华佗不仅治病，更发明了五禽戏强身健体——预防胜于治疗。

Submit your AI agent code for a Six-Meridian Security Checkup（六经脉安全体检）:

🛡️ Meridian 1: Sandbox Isolation     → 30/100 ⚠️ CRITICAL
🔑 Meridian 2: Key Management        → 85/100 ✅ Healthy
🧠 Meridian 3: Prompt Security       → 45/100 ⚠️ At risk
🔒 Meridian 4: Output Sanitization   → 60/100 🟡 Needs care
⏱️ Meridian 5: Resilience Design     → 72/100 🔵 Good
📊 Meridian 6: Observability         → 55/100 🟡 Needs care

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Overall Health Score: 58/100  🟡 Needs Nourishing
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

💊 Top Nourishing Prescriptions:
  1. 🏭 MCP & Skills Supply Chain Audit (100% Industrial Grade Sigstore/OSV/LLM) ✨ NEW
  2. Add execution sandboxing (RestrictedPython / Docker)
  3. Implement prompt injection defenses
  4. Add structured logging and tracing

Your AI is running naked in production. Do you know it?

🦠 Epidemic Alert — The AI Pandemic Nobody Talks About

疫情通报 — 没有人谈论的 AI 大流行

Every day, thousands of AI systems go into production already infected:

🔓 No sandbox — agent code executes with full system privileges
🔑 Hardcoded secrets — API keys sitting in plain text, spreading like a virus
🧠 No prompt defense — one injection away from a full outbreak
📊 No observability — zero symptom monitoring, blind until collapse

This is not a future risk. This is an active pandemic.

And when things break? Developers spend hours digging through GitHub Issues, Discord threads, Reddit posts, and half-abandoned blog articles.

The cure exists. Buried under seven layers of irrelevant search results.

Stack Overflow waits for someone to answer. ChatGPT might hallucinate a treatment. GitHub Issues drown in duplicate reports.

CyberHuaTuo is the WHO of the AI world — tracking the epidemic, prescribing battle-tested cures, and vaccinating before the next outbreak.

🤔 Why CyberHuaTuo?

	🩺 CyberHuaTuo	🔍 Stack Overflow	🤖 ChatGPT	📋 GitHub Issues
AI-specific knowledge	✅ Purpose-built	⚠️ General	⚠️ Generic	⚠️ Scattered
Version-aware diagnosis	✅ Auto-detect	❌ Manual	❌ Outdated cutoff	❌ Manual
Structured prescriptions	✅ Root cause + fix + verify	⚠️ Varies	⚠️ May hallucinate	⚠️ Varies
Cross-framework mapping	✅ LangChain ↔ LlamaIndex ↔ CrewAI	❌ Siloed	⚠️ Inconsistent	❌ Siloed
Security health check	✅ Six-meridian scan	❌ None	❌ None	❌ None
Preventive nourishing	✅ Proactive hardening	❌ Reactive only	❌ Reactive only	❌ Reactive only
Official doc retrieval	✅ Context7 real-time docs	❌ None	❌ Outdated	❌ None
Speed to cure	⚡ Seconds	🕐 Hours	⚡ Fast but risky	🕐 Hours

🔮 How It Works

望闻问切 — The 2,000-Year-Old Diagnostic Protocol

Inspired by traditional Chinese medicine's four-step diagnosis, refined for the AI age:

  Your Error / Your Code
       │
       ▼
  ┌──────────┐
  │  望 Look  │──→ Parse stack trace, detect framework, version & environment
  └─────┬────┘
       ▼
  ┌───────────┐
  │ 闻 Listen  │──→ Correlate with known issues, breaking changes & CVEs
  └──────┬────┘
       ▼
  ┌─────────┐
  │  问 Ask  │──→ Smart follow-up (only when critical info is missing)
  └────┬────┘
       ▼
  ┌────────────┐
  │ 切 Diagnose │──→ Semantic search → LLM reasoning → Ranked prescriptions
  └──────┬─────┘
       ▼
  💊 Prescriptions (ranked by cure rate)
  🌿 Nourishing advice (preventive recommendations)

Every prescription includes:

🎯 Root cause — not just what to do, but why it broke
🔧 Copy-paste fix — step-by-step with ready-to-run code
📌 Version-locked — verified for your exact framework version
🔄 Cross-framework map — "LangChain issue? Here's the LlamaIndex equivalent."
📚 Official docs — pulled from framework's latest documentation via Context7
✅ Community-verified — tested, upvoted, and marked "Cured ✅"

📋 What We Treat — Not Just Agents

If it has AI in it, CyberHuaTuo sees it. No framework left behind.

🤖 AI Agent Frameworks

Framework	Cases	Status	Join 👇
LangChain	10+	🟢 Live	Add a case →
MCP (Anthropic)	5+	🟢 Live	Add a case →
CrewAI	5+	🟢 Live	Add a case →
LlamaIndex	—	🟡 Accepting PRs	Be the first →
OpenAI Agents SDK	—	🟡 Accepting PRs	Be the first →
AutoGen	—	🟡 Accepting PRs	Be the first →
DSPy	—	🔵 Planned	Vote →

🧠 AI / ML / Deep Learning

Technology	Cases	Status	Join 👇
PyTorch	—	🟡 Accepting PRs	Be the first →
Transformers (HuggingFace)	—	🟡 Accepting PRs	Be the first →
TensorFlow	—	🟡 Accepting PRs	Be the first →

🏗️ Platform & Custom Agents

Platform	Cases	Status	Join 👇
GPTs / Coze / Dify	—	🟡 Accepting PRs	Be the first →
Custom-built Agents	—	🟡 Accepting PRs	Be the first →

🌿 Nourishing Prescriptions (养生药方)

Category	Cases	Status	Join 👇
🛡️ Security Sandbox	2+	🟢 Live	Add →
🔒 Security Hardening	3+	🟢 Live	Add →
🔗 Supply Chain Audit	1	🟢 Live	Add →
⚡ Performance Tuning	—	🟡 Accepting PRs	Be the first →

Every framework starts with one case. Every nourishing recipe starts with one insight. Yours could be the first.

🚀 Quick Start

Prerequisites

Python 3.10+
(Optional) LLM API Key for AI-powered diagnosis (OpenAI / Anthropic / DeepSeek / Gemini / Ollama)

Option A: One-Click Launch

git clone https://github.com/JinNing6/CyberHuaTuo-Plugin.git
cd CyberHuaTuo

# Windows
start.bat

# macOS / Linux
chmod +x start.sh && ./start.sh

Auto-installs dependencies and opens the Web UI at http://127.0.0.1:8000

Option B: Manual Setup

git clone https://github.com/JinNing6/CyberHuaTuo-Plugin.git
cd CyberHuaTuo

pip install -r requirements.txt

# (Optional) Configure LLM for AI diagnosis
cp .env.example .env
# Edit .env — add your API Key

python -m cyberhuatuo serve
# → Browser auto-opens http://127.0.0.1:8000

CLI Commands — Two Modes, One Engine

CyberHuaTuo supports two access modes — pick the one that fits your workflow:

Mode	Entry Point	For
⌨️ CLI	`cyberhuatuo <command>`	Developers using terminal directly
🔌 MCP	`cyberhuatuo-mcp`	AI editors (Cursor / Claude / VS Code)

# 🩺 Core Diagnosis / 核心诊断
cyberhuatuo diagnose "ImportError: cannot import 'ChatOpenAI'"  # AI 诊断
cyberhuatuo search "CUDA out of memory" --framework pytorch     # 搜索知识库
cyberhuatuo checkup --file my_agent.py                          # 安全体检
cyberhuatuo docs langchain "How to configure RAG pipeline"      # 官方文档检索

# 📊 Knowledge & Analytics / 知识与分析
cyberhuatuo stats                            # 知识库统计
cyberhuatuo frameworks --search pytorch      # 支持框架列表
cyberhuatuo taxonomy list                    # CHT 根因编码系统
cyberhuatuo trends --framework langchain     # CHT 趋势分析
cyberhuatuo digest                           # 本周药方摘要
cyberhuatuo epidemic --action check          # 疫情预警

# 🧬 Alchemist System / 炼丹师体系
cyberhuatuo challenge --username your-github-username --framework langchain
cyberhuatuo mission --username your-github-username --framework langchain --sect Azure-Sect --members your-github-username friend-github-username
cyberhuatuo bounty --username your-github-username --framework auto --top-n 8 --release-tag v0.2.5 --target-contributors 3
cyberhuatuo launch --username your-github-username --framework langchain --release-tag v0.2.5
cyberhuatuo launch-campaign --username your-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3
cyberhuatuo traction-proof --username your-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3
cyberhuatuo traction-proof --username your-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3 --record-snapshot
python -m cyberhuatuo candidate-install-smoke --username your-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3
cyberhuatuo market-copy --username your-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3
cyberhuatuo record-market --username your-github-username --framework langchain --channel pypi --status submitted --submission-url <reviewable public URL> --release-tag v0.2.5
cyberhuatuo market-status --username your-github-username --framework langchain --release-tag v0.2.5
cyberhuatuo first-invite --username your-github-username --invitee external-contributor-github-username --framework langchain --release-tag v0.2.5 --target-contributors 3 --source-url https://github.com/JinNing6/CyberHuaTuo-Plugin/issues/123
cyberhuatuo record-return --username your-github-username --framework langchain --surface "PyPI release" --source-url https://example.com/post
cyberhuatuo activation --username your-github-username --framework langchain --sect Azure-Sect --members your-github-username friend-github-username --top-n 10
cyberhuatuo flywheel --username your-github-username --framework langchain --sect Azure-Sect --members your-github-username friend-github-username --top-n 10
cyberhuatuo record-share --username your-github-username --framework langchain --share-url https://example.com/share
cyberhuatuo share-report --username your-github-username --framework langchain --top-n 10
cyberhuatuo share-leaderboard --framework langchain --top-n 10
cyberhuatuo ladder your-github-username --framework langchain
cyberhuatuo ranking your-github-username     # 查看个人排名
cyberhuatuo leaderboard                     # 全球封神榜
cyberhuatuo card your-github-username        # 生成分享卡片
cyberhuatuo badge your-github-username       # 生成 GitHub Profile 魂环徽章
cyberhuatuo quest your-github-username --framework langchain
cyberhuatuo campaign your-github-username --framework langchain
cyberhuatuo duel your-github-username friend-github-username --framework langchain
cyberhuatuo mentor mentor-github apprentice-github --framework langchain
# GitHub mentor pact IssueOps: .github/ISSUE_TEMPLATE/soul-ring-mentor.yml
# GitHub mentor pact workflow: .github/workflows/soul-ring-mentor.yml
cyberhuatuo tournament alice bob carol dave --framework langchain --event Agent-Cup
cyberhuatuo tournament-settle alice bob carol dave --framework langchain --event Agent-Cup
# GitHub tournament IssueOps: .github/ISSUE_TEMPLATE/soul-ring-tournament.yml
# GitHub tournament workflow: .github/workflows/soul-ring-tournament.yml
# GitHub sect recruitment IssueOps: .github/ISSUE_TEMPLATE/soul-ring-sect-recruit.yml
# GitHub sect recruitment workflow: .github/workflows/soul-ring-sect-recruit.yml
cyberhuatuo arena your-github-username --top-n 10
cyberhuatuo season --framework langchain --top-n 10
# GitHub season board IssueOps: .github/ISSUE_TEMPLATE/soul-ring-season.yml
# GitHub season board workflow: .github/workflows/soul-ring-season.yml
# GitHub growth flywheel IssueOps: .github/ISSUE_TEMPLATE/soul-ring-growth-flywheel.yml
# GitHub growth flywheel workflow: .github/workflows/soul-ring-growth-flywheel.yml
# GitHub bounty board IssueOps: .github/ISSUE_TEMPLATE/soul-ring-bounty-board.yml
# GitHub bounty board workflow: .github/workflows/soul-ring-bounty-board.yml
# GitHub launch campaign IssueOps: .github/ISSUE_TEMPLATE/soul-ring-launch-campaign.yml
# GitHub launch campaign workflow: .github/workflows/soul-ring-launch-campaign.yml
# Public traction proof: cyberhuatuo traction-proof reads GitHub REST API, GitHub Pull Requests API, GitHub Contents API, GitHub Releases API, PyPI JSON API, and the local activation/share ledger.
# Readiness gates: PyPI latest version must not lag local growth tools, and default-branch IssueOps forms/workflows must exist before issues/new?... links count as live acquisition loops.
# Target contributor progress uses real issue/PR/ledger identities; PRs stay separate from IssueOps counts, and stars, forks, watchers, and downloads are not used as contributors.
# Snapshot history is opt-in: add --record-snapshot to append an append-only real JSONL snapshot and compare velocity deltas.
cyberhuatuo sect Azure-Sect your-github-username friend-github-username --framework langchain
cyberhuatuo sect-recruit Azure-Sect your-github-username friend-github-username --invitee new-member-github --framework langchain
cyberhuatuo sect-quest Azure-Sect your-github-username friend-github-username --framework langchain
cyberhuatuo sect-hall Azure-Sect your-github-username friend-github-username --framework langchain
cyberhuatuo sect-duel Azure-Sect Shadow-Sect --challenger-members your-github-username friend-github-username --rival-members rival-a rival-b --framework langchain
cyberhuatuo sect-arena --sect Azure-Sect your-github-username friend-github-username --sect Shadow-Sect rival-a rival-b --framework langchain
cyberhuatuo record                          # 诊疗档案
cyberhuatuo subscribe subscribe -f langchain # 订阅框架更新

# ⛏️ Contribution / 知识贡献
cyberhuatuo save --title "Fix" --prescription "..." --framework langchain
cyberhuatuo upload --title "Fix" --prescription "..." --framework langchain --contributor your-github-username
cyberhuatuo mine search --repo langchain-ai/langchain

# 🚀 Server / 服务启动
cyberhuatuo serve                            # 启动 Web 诊断服务
cyberhuatuo serve --port 9000 --reload       # 自定义端口 + 开发模式
cyberhuatuo rebuild                          # 重建向量索引

Every MCP tool has a CLI counterpart. Same engine, same knowledge base, different interface.

每个 MCP 工具都有对应的 CLI 命令。 同一个引擎，同一个知识库，不同的界面。

Configuration

Copy .env.example to .env:

# LLM Provider (choose one)
OPENAI_API_KEY=sk-your-key              # OpenAI
ANTHROPIC_API_KEY=sk-ant-your-key       # Anthropic
DEEPSEEK_API_KEY=sk-your-key            # DeepSeek
GEMINI_API_KEY=your-key                 # Google Gemini
OLLAMA_BASE_URL=http://localhost:11434  # Ollama (local, free)

# Diagnosis model
DIAGNOSIS_MODEL=gpt-4o-mini   # or claude-sonnet-4-20250514 / deepseek-chat / gemini-1.5-pro

# Server
PORT=8000

No API Key? No problem. Vector search works without any key. AI-powered 望闻问切 diagnosis and 🌿 nourishing checkup require one of the above.

📊 Epidemic Report — Live Intelligence

A living dashboard tracking what's breaking across the AI ecosystem right now:

🔥 Trending issues and their cure rates
🗺️ Heatmap of problems across frameworks
📈 Framework health scores — know before you adopt
⛏️ Issue mining — real-time extraction from GitHub

🛠️ Tech Stack

Layer	Technology
CLI	Python argparse (marketplace, MCP, and soul-ring subcommands; zero extra dependencies)
MCP Server	FastMCP (stdio transport for AI editor integration)
Backend	Python · FastAPI · Uvicorn
Frontend	Jinja2 Templates (server-rendered)
Vector DB	ChromaDB (embedded, zero-config)
Ephemeral Layer	GitHub Issues API (dual-layer prescription search)
LLM Gateway	LiteLLM (OpenAI / Anthropic / DeepSeek / Gemini / Ollama)
Doc Retrieval	Context7 API (real-time official framework docs)
Security Engine	Six-Meridian Audit (AI-powered code analysis)
GitHub Bot	GitHub Actions + lightweight text matching
CI/CD	GitHub Actions (auto-promotion, validation, labeling)

🤖 GitHub Bot — Zero-Friction Diagnosis

The greatest doctor doesn't wait in the clinic — they meet patients where they are.

大医不坐堂候诊——而是行走于民间，哪里有病痛，哪里就有华佗。

Install the CyberHuaTuo Bot on any repo — automatic prescription recommendations on every Issue.

┌─────────────────────────────────────────────┐
│  New Issue opened                           │
│  "ImportError: cannot import ChatOpenAI"    │
│                                             │
│  💬 @CyberHuaTuo Bot replies:              │
│  ┌─────────────────────────────────────┐    │
│  │ 🩺 赛博华佗 · 自动诊断             │    │
│  │                                     │    │
│  │ 💊 药方 1 (92% match):             │    │
│  │ pip install langchain-openai        │    │
│  │ from langchain_openai import ...    │    │
│  │                                     │    │
│  │ 📋 Full prescription →              │    │
│  └─────────────────────────────────────┘    │
└─────────────────────────────────────────────┘

How it works

Trigger	Action
🆕 New Issue opened	Auto-match prescriptions from the knowledge base
💬 `@CyberHuaTuo` in any comment	On-demand diagnosis

Install on your repo (3 minutes)

# Copy the Bot workflow to your repo
mkdir -p .github/workflows
curl -o .github/workflows/bot-prescribe.yml \
  https://raw.githubusercontent.com/JinNing6/CyberHuaTuo-Plugin/main/.github/workflows/bot-prescribe.yml

📖 Full installation guide →

The Flywheel

  Developers see Bot is useful → Contribute more prescriptions
        → Bot becomes more accurate → More repos install Bot
            → CyberHuaTuo brand appears before thousands of developers daily

Imagine: If LangChain's official repo installs this Bot, every new Issue gets a prescription recommendation. CyberHuaTuo's brand appears before thousands of developers every day. 🚀

⌨️ CLI — Terminal Power at Your Fingertips

The terminal is the alchemist's cauldron — raw, powerful, and always ready.

终端就是炼丹师的鼎炉——原始、强大、随时待命。

CyberHuaTuo ships a full-featured CLI with marketplace, MCP, and soul-ring subcommands — every MCP tool has a terminal counterpart. No AI editor required. No GUI needed. Just type and heal.

赛博华佗提供完整的 CLI 命令行界面，包含 21 个子命令——每个 MCP 工具都有对应的终端命令。无需 AI 编辑器，无需图形界面，敲命令，治百病。

# Install
python -m pip install --upgrade "cyberhuatuo @ git+https://github.com/JinNing6/CyberHuaTuo-Plugin.git@v0.2.5"
python -m pip install --upgrade cyberhuatuo  # after PyPI latest matches v0.2.5

# Or run from source
python -m cyberhuatuo --help

🎬 Terminal Animations — 终端动画特效

⌨️ Available CLI Commands

Command	Description
🩺 Core Diagnosis / 核心诊断
`diagnose <query>`	🩺 望闻问切 AI 诊断 — paste your error, get a cure
`search <query>`	🔍 双层语义搜索 (ChromaDB + GitHub Issues)
`checkup --file <path>`	🛡️ 六经脉安全体检 — audit your Agent code
`docs <framework> <query>`	📚 实时官方文档检索 (50+ frameworks via Context7)
📊 Analytics / 知识分析
`stats`	📦 知识库统计 — case counts by framework
`frameworks`	📋 支持框架列表 — filterable by category / keyword
`taxonomy <action>`	🧬 CHT 根因编码 — list, lookup, classify
`trends`	📈 CHT 趋势分析 — heatmaps & surge alerts
`digest`	📰 本周药方摘要 — weekly prescription digest
`epidemic`	🦠 疫情预警 — framework health monitoring
🧬 Alchemist System / 炼丹师体系
`challenge --username <username> --framework <framework>`	🔮 first Soul Ring onramp — submit, rank, share
`mission --username <username> --framework <framework> --sect <sect-name> --members <members...>`	🔮 Soul Ring mission hall — one-screen Issue, PR, personal ring, and sect/team onboarding
`bounty --username <username> --framework auto --top-n <N> --release-tag <tag> --target-contributors <N>`	Soul Ring Bounty Board from real local framework coverage gaps, claimable First Soul Ring Issue routes, and no fake rewards
`launch --username <username> --framework <framework> --release-tag <tag>`	Soul Ring launch scroll for PyPI, Claude, Codex, MCP, and First Soul Ring routing
`launch-campaign --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	Soul Ring Launch Campaign for turning PyPI, Claude, and Codex attention into a target first-ring contributor loop with recap and next sprint
`traction-proof --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	Soul Ring Traction Proof from GitHub REST API, GitHub Pull Requests API, GitHub Contents API IssueOps readiness, GitHub Releases API release.published readiness, PyPI JSON API package readiness, and local ledger identities
`traction-proof --username <username> --framework <framework> --release-tag <tag> --target-contributors <N> --record-snapshot`	Opt-in append-only Soul Ring Traction Proof history with velocity deltas from prior real snapshots
`cyberhuatuo install-command --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	CyberHuaTuo Install Command: fetches real PyPI JSON API latest-version proof, recommends canonical `python -m pip install --upgrade cyberhuatuo` only when the registry is current, otherwise prints the bounded Git Tag Candidate Install Bridge plus `candidate-install-smoke`, then routes to `challenge`, `proof-pack`, `market-copy`, and `traction-proof`; MCP equivalent: `current_install_command`
`python -m cyberhuatuo candidate-install-smoke --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	Candidate Install Smoke Gate: creates a disposable venv, installs the exact public Git tag with pip Direct URL syntax, verifies installed version, console command, Install Decision Surface, and proof/invite route, cleans up on success, and retains the temp dir on failure; it is not run automatically inside lightweight CI and does not close the PyPI install loop
`first-invite --username <username> --invitee <github> --framework <framework> --release-tag <tag> --target-contributors <N> --source-url <https-url>`	First Contributor Invite Pack for targeting one external contributor with First Soul Ring issue URLs, record-session / challenge commands, proof rechecks, and no fake progress
`market-ready --no-remote` / `market-ready --remote --strict-remote`	Marketplace Readiness Gate with a first-screen Flywheel Closure Verdict (`closed`, `not closed`, or `unverified`), Ready gates / total gate counts, a Launch Closure Checklist, and a First Public Proof Kit for PyPI, Claude, Codex, MCP, version sync, IssueOps files, release trigger or protected publish fallback, registry proof, a Git Tag Candidate Install Bridge for stale PyPI recovery, public launch blockers, proof Issue URLs, Community Challenge Pack, ledger commands, a Public Release Operator Runbook, and copy-ready proof copy
`launch-assets [--username <username> --framework <framework> --release-tag <tag> --target-contributors <N>]`	Local Launch Asset Audit for issue forms, comment-only workflows, package metadata, plugin manifests, MCPB assets, a Full Public Growth Release Bundle, Dirty Worktree Release Coverage, and exact review-only `git add` commands. When release context is supplied, the Public Release Operator Runbook preserves the same release/user/target context; it does not stage files, publish releases, upload to PyPI, mutate remotes, or claim traction
`proof-pack --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	No-Network First Public Proof Pack with Growth/Share/Bounty Issue URLs, created-Issue placeholders, Community Challenge Pack for Tournament/Mentor/Sect/Season routes, a Protected Publish Fallback command, Git Tag Candidate Install Bridge plus Candidate Install Smoke Gate, Install Decision Surface / `current_install_command` route, Growth and Bounty `record-return` CLI ledger commands, Share `record-share` attribution, an External Contributor Path with pasted Recommended Install / first-session command / first contribution command / Share Proof Issue URL / contributor-counting rule, recheck commands, and copy-ready proof text; it does not fetch public metrics, close the PyPI install loop, or invent traction
`market-copy --username <username> --framework <framework> --release-tag <tag> --target-contributors <N>`	Marketplace Submission Copy Pack with PyPI listing copy, Claude MCPB listing copy, Codex plugin listing copy, GitHub Release post, public proof CTA, Community Challenge Pack, Git Tag Candidate Install Bridge plus Candidate Install Smoke Gate, project URLs, install/validation commands, Marketplace Submission Ledger commands, and non-fabrication rules for market submission
`record-market --username <username> --framework <framework> --channel <channel> --status <status> --submission-url <https-url> --release-tag <tag>`	Record a reviewable public URL for PyPI, Claude Code, Claude Desktop MCPB, Codex, GitHub Release, or agent-marketplace submission evidence into the local Marketplace Submission Ledger
`market-status --username <username> --framework <framework> --release-tag <tag>`	Report latest Marketplace Submission Ledger status by channel with missing record-market commands and traction-proof rechecks
`flywheel --username <username> --framework <framework> --sect <sect-name> --members <members...>`	Soul Ring Growth Flywheel with real bottleneck, next commands, and missing metric disclosure
`ladder <username> --framework <framework>`	🔮 Soul Ring breakthrough ladder — real next gate, full thresholds, and copy-ready chase commands
`evidence <username> --framework <framework> --amount <N> --source-url <https-url>`	Soul Ring Evidence Card with append-only reviewable public evidence for high-realm gates; progress, ranks, rewards, downloads, and contributors are not invented
`ranking <username>`	🏆 个人排名 — title, rank & soul rings
`leaderboard`	🏆 全球封神榜 — top alchemists ranked
`card <username>`	📋 分享卡片 — cyberpunk cultivation archive
`badge <username>`	🔮 GitHub Profile badge kit — pasteable soul-ring badge
`visual <username> --framework <framework> --output-dir <dir>`	Soul Ring Visual Artifact — writes a chat-visible animated GIF plus PNG fallback and returns Markdown image embeds for Codex, Claude, and other agent chats
`quest <username> --framework <framework>`	🔮 Soul Ring quest board — mine, upload, share
`campaign <username> --framework <framework>`	🔮 Soul Ring campaign pack — GitHub Profile, X/Weibo, PR comment & agent prompt
`duel <challenger> <rival> --framework <framework>`	🔮 Soul Ring duel card — invite a real GitHub user into a verified contribution race
`mentor <mentor> <apprentice> --framework <framework>`	🔮 Soul Ring mentor pact — turn a senior contributor into a first-ring onboarding guide
`arena <username> --top-n <N>`	🔮 Soul Ring arena snapshot — shareable real leaderboard with your next rival
`season --framework <framework> --top-n <N>`	🔮 Soul Ring season board — current real leaderboard event pack with champion, next chase, and share copy
`sect <sect-name> <members...> --framework <framework>`	🔮 Soul Ring sect card — aggregate real GitHub members into a shareable team banner
`sect-recruit <sect-name> <members...> --invitee <github> --framework <framework>`	🔮 Soul Ring sect recruitment scroll — invite one real GitHub user with an admission trial and join command
`sect-quest <sect-name> <members...> --framework <framework>`	🔮 Soul Ring sect quest board — assign real next actions across team members
`sect-hall <sect-name> <members...> --framework <framework>`	🔮 Soul Ring sect hall — assign Outer / Inner / Core / Deacon / Elder posts from real member counts
`sect-duel <challenger-sect> <rival-sect> --challenger-members <members...> --rival-members <members...> --framework <framework>`	🔮 Soul Ring sect duel card — compare two real GitHub teams and invite a team challenge
`sect-arena --sect <sect-name> <members...> --sect <sect-name> <members...> --framework <framework>`	🔮 Soul Ring sect arena snapshot — rank multiple real GitHub teams and publish the next chase
`record`	📋 诊疗档案 — diagnosis history & follow-ups
`subscribe`	📬 框架订阅 — get notified about new prescriptions
⛏️ Contribution / 知识贡献
`save`	📥 保存药方到本地
`upload`	🌐 上传药方到 GitHub (需 GITHUB_TOKEN)
`mine search/batch`	⛏️ GitHub Issues 淘金提炼
🚀 Server / 服务
`serve`	🚀 启动 Web 诊断服务
`rebuild`	🔄 重建向量索引

Soul Ring Visual Route

Run cyberhuatuo visual <username> --framework <framework> in the CLI or call MCP soul_ring_visual_artifact to generate a Markdown-rendered GIF plus PNG fallback. The artifact uses current real contribution data and does not invent ranks, downloads, retention, referrals, or rewards.

💡 Tip: Run cyberhuatuo <command> --help for detailed usage of any command.

💡 提示: 运行 cyberhuatuo <命令> --help 查看任何命令的详细用法。

🔌 MCP Server — AI Editor Integration

Install once. Every AI editor becomes a CyberHuaTuo clinic.

一次安装，所有 AI 编辑器都变成赛博华佗诊所。

⚡ Install — One Command

python -m pip install --upgrade "cyberhuatuo @ git+https://github.com/JinNing6/CyberHuaTuo-Plugin.git@v0.2.5"
python -m pip install --upgrade cyberhuatuo  # after PyPI latest matches v0.2.5

Candidate Git tag install should pass python -m cyberhuatuo candidate-install-smoke before outside contributor invites; after PyPI latest-version proof is current, use the registry command. Brand Matrix aliases are for the registry-current path only.

🔗 Connect to Your AI Editor

Add this to your AI tool's MCP config (Claude Desktop / Cursor / VS Code / Gemini CLI):

{
  "mcpServers": {
    "cyberhuatuo": {
      "command": "uvx",
      "args": ["--from", "cyberhuatuo", "cyberhuatuo-mcp"]
    }
  }
}

uvx auto-installs from PyPI and starts the MCP Server — zero manual setup. For GitHub latest dev version, use: "args": ["--from", "git+https://github.com/JinNing6/CyberHuaTuo-Plugin", "cyberhuatuo-mcp"]

🧩 Plugin Manifests — Codex + Claude Code

This repository is plugin-ready for both major coding-agent hosts:

Codex: .codex-plugin/plugin.json points to ./skills/ and ./.mcp.json.
Claude Code: .claude-plugin/plugin.json points to the same ./skills/ and ./.mcp.json.
Codex marketplace catalog: .agents/plugins/marketplace.json lets users add this GitHub repository as a Codex plugin marketplace.
Claude Code marketplace catalog: .claude-plugin/marketplace.json lets users add this GitHub repository as a Claude Code plugin marketplace.

Marketplace install:

claude plugin marketplace add JinNing6/CyberHuaTuo-Plugin
claude plugin install cyberhuatuo-plugin@cyberhuatuo

codex plugin marketplace add JinNing6/CyberHuaTuo-Plugin

For local Claude Code testing from the repository root:

claude --plugin-dir .

🚀 Marketplace Release Path

The release plan is tracked in docs/MARKETPLACE_RELEASE.md:

PyPI Trusted Publishing: cyberhuatuo already exists on PyPI, so this repository must be added as an additional Trusted Publisher, then release a version newer than 0.1.0 through .github/workflows/publish-pypi.yml. The workflow supports release.published and a protected manual workflow_dispatch release_tag fallback; both paths require an existing v* tag, verify the tag is reachable from origin/main, verify the package version matches the tag, and publish with OIDC instead of PYPI_TOKEN.
Claude markets: ship Claude Code through .claude-plugin/plugin.json and .claude-plugin/marketplace.json, submit to Anthropic's official directory after claude plugin validate ., then package Claude Desktop through claude-desktop/manifest.json plus .github/workflows/package-claude-mcpb.yml for MCPB / Connectors Directory submission.
Codex plugin directory: keep .codex-plugin/plugin.json, .agents/plugins/marketplace.json, .mcp.json, and the PyPI uvx --from cyberhuatuo cyberhuatuo-mcp entrypoint stable for workspace rollout.

🩺 Available MCP Tools

Tool	Description
🩺 Core Diagnosis / 核心诊断
`diagnose`	🩺 望闻问切 AI diagnosis — searches both 📜 permanent expert + ⚡ temporary intern prescriptions
`search_knowledge_base`	🔍 Dual-layer semantic search (ChromaDB + GitHub Issues)
`security_checkup`	🛡️ Six-meridian security audit on your Agent code
`fetch_official_docs`	📚 Real-time official docs via Context7 (50+ frameworks)
⛏️ Knowledge & Contribution / 知识贡献
`mine_github_issue`	⛏️ Extract & refine GitHub Issues into standard cases
`save_prescription`	📥 Save prescriptions locally + auto-sync to GitHub
`upload_prescription`	🌐 Upload prescriptions — auto-creates GitHub Issue (⚡ temporary intern), CI promotes to 📜 permanent expert
`prescription_eval`