Skillget

Clade

mcp
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 5 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose

This MCP server transforms Claude Code into an autonomous coding agent, equipping it with 29 skills, 5 agents, and a safety guardian so it can execute tasks unattended.

Security Assessment

The code scan of 12 files found no dangerous patterns or hardcoded secrets, and the tool does not request broadly dangerous permissions. However, by design, it executes local shell commands (via `claude -p` and bash hooks) and interacts directly with your local file system and git repositories. A built-in safety hook attempts to block risky operations like `rm -rf`, `DROP TABLE`, and force pushes, but giving any AI autonomy over local execution inherently elevates risk. Overall risk is rated as Medium because autonomous shell execution requires the user to place significant trust in the tool's internal guardrails.

Quality Assessment

The project is actively maintained (last updated today) and properly licensed under the permissive MIT license. The repository includes detailed, bilingual documentation with clear installation instructions and a well-documented architecture. Community trust is currently low, as it is a new or niche project with only 5 GitHub stars and limited external visibility.

Verdict

Use with caution: the code is transparent and actively maintained, but its core feature of autonomous shell execution means you should carefully review the safety hooks before running it in any critical environments.
SUMMARY

Turn Claude Code from a chat assistant into an autonomous coding system

README.md

English | 中文

Clade

PyPI PRs Welcome good first issue

Clade

Autonomous coding, evolved.

29 skills, 14 hooks, 5 agents, a safety guardian, and a correction learning loop — all working together so Claude codes better, catches its own mistakes, and can run unattended overnight while you sleep.

If this saves you time, a star helps others find it. Something broken? Open an issue.

Blog post: Building Clade — motivation, design decisions, and lessons learned.

Table of Contents

  1. Install
  2. MCP Server
  3. What It Does
  4. Skills
  5. Hooks
  6. Supported Languages
  7. Documentation
  8. Repo Structure
  9. Contributing
  10. License

Install

Full Framework (recommended)

git clone https://github.com/shenxingy/clade.git
cd clade && ./install.sh

Installs skills, hooks, agents, scripts, and safety guardian. Start a new Claude Code session to activate.

Requires: jq. Platform: Linux and macOS.

MCP Server Only

If you just want the skills in Cursor, Windsurf, Claude Desktop, or any MCP client:

pip install clade-mcp

See MCP Server below for configuration.

MCP Server — Use Skills in Any AI Editor

The MCP server exposes all 29 Clade skills as callable tools via the Model Context Protocol. Works with any MCP-compatible client.

Claude Desktop / Claude Code:

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

Cursor / Windsurf:

{
  "mcpServers": {
    "clade": { "command": "clade-mcp" }
  }
}

Prerequisite: Claude Code CLI must be installed — skills execute via claude -p.

What It Does

When What fires Effect
Session opens in a git repo session-context.sh Loads git context, handoff state, correction rules, model guidance
Claude runs a bash command pre-tool-guardian.sh Blocks dangerous ops: migrations, rm -rf, force push, DROP TABLE
Claude edits code post-edit-check.sh Async type-check (tsc, pyright, cargo check, go vet, etc.)
You correct Claude correction-detector.sh Logs correction, prompts Claude to save a reusable rule
Claude marks task done verify-task-completed.sh Adaptive quality gate: compile + lint, build + test in strict mode

See How It Works for the full hook reference (14 hooks).

Skills (29)

Core Workflow

Skill What it does
/commit Split changes into logical commits by module, push by default
/sync Check off completed TODOs, append session summary to PROGRESS.md
/review 8-phase coverage review — finds AND fixes issues, loops until clean
/verify Verify project behavior anchors (compile, test, lint)

Autonomous Operation

Skill What it does
/start Autonomous session launcher — morning brief, overnight runs, cross-project patrol
/loop GOAL Goal-driven improvement loop — supervisor plans, workers execute in parallel
/batch-tasks Execute TODO steps via unattended sessions (serial or parallel)
/orchestrate Decompose goals into tasks for worker execution
/handoff Save session state for context relay between agents
/pickup Resume from previous handoff — zero-friction restart
/worktree Create git worktrees for parallel sessions

Code Quality

Skill What it does
/review-pr N AI code review on a PR diff — Critical / Warning / Suggestion
/merge-pr N Squash-merge PR and clean up branch
/investigate Root cause analysis — no fix without confirmed hypothesis
/incident DESC Incident response — diagnose, postmortem, follow-up tasks
/cso Security audit (OWASP + STRIDE)
/map Generate ARCHITECTURE.md with module graph + file ownership

Research & Planning

Skill What it does
/research TOPIC Deep web research, synthesize to docs/research/
/model-research Latest Claude model data + auto-update configs
/next Multi-angle priority session — surface best next move
/brief Morning briefing — overnight commits, costs, next steps
/retro Engineering retrospective from git history
/frontend-design Create production-grade frontend interfaces

System

Skill What it does
/audit Clean up correction rules — promote, deduplicate, remove stale
/document-release Post-ship doc sync (README, CHANGELOG, CLAUDE.md)
/pipeline Health check for background pipelines
/provider Switch LLM provider
slt Toggle statusline quota pace indicator

See When to Use What for detailed usage guidance.

Supported Languages

Auto-detected — hooks and agents adapt to your project:

Language Edit check Type checker Test runner
TypeScript / JavaScript tsc (monorepo-aware) tsc jest / vitest
Python pyright / mypy pyright / mypy pytest
Rust cargo check cargo check cargo test
Go go vet go vet go test
Swift / iOS swift build swift build swift test
Kotlin / Android / Java gradlew gradlew gradle test
LaTeX chktex chktex —

All checks are opt-in by detection — if the tool isn't installed, the hook silently skips.

Documentation

Guide Contents
Maximize Throughput Skip permissions, batch tasks, parallel worktrees, terminal + voice
Orchestrator Web UI Chat-to-plan, worker dashboard, settings, iteration loop
Overnight Operation Task queue, parallel sessions, context relay, safety
How It Works Hooks, agents, skills internals, correction learning, model selection
Configuration Settings, thresholds, adding custom hooks/agents/skills
When to Use What Detailed usage guidance for every skill

Dotfile Sync

Keep ~/.claude/ in sync across machines — memory, corrections, skills, hooks, and scripts.

~/.claude/scripts/sync-setup.sh            # auto-detect NFS or GitHub
~/.claude/scripts/sync-setup.sh --github   # explicit GitHub backend

Fully automatic once configured. See Configuration for details.

Repo Structure

clade/
├── install.sh               # One-command deployment
├── uninstall.sh             # Clean removal
├── mcp-package/             # PyPI package (clade-mcp)
├── orchestrator/            # FastAPI web UI + worker pool + task queue
│   ├── server.py            # App, routes, WebSocket
│   ├── worker.py            # WorkerPool, SwarmManager
│   ├── task_queue.py        # SQLite-backed task CRUD
│   ├── mcp_server.py        # MCP server (local dev version)
│   └── web/                 # Single-page dashboard
├── configs/
│   ├── skills/              # 29 skill definitions (SKILL.md + prompt.md)
│   ├── hooks/               # 14 event hooks + lib/
│   ├── agents/              # 5 agent definitions
│   └── scripts/             # 30 shell + Python utilities
├── adapters/openclaw/       # OpenClaw integration (mobile monitoring)
├── templates/               # Settings, CLAUDE.md, corrections templates
└── docs/                    # Guides and research

OpenClaw Integration

Monitor and control overnight loops from your phone via OpenClaw.

Skill Trigger Effect
clade-status "how's the loop going" Iteration progress, cost, commits
clade-control "start a loop to fix tests" Start/stop autonomous loops
clade-report "what did it do overnight" Session report, cost breakdown

See adapters/openclaw/README.md for setup.

Contributing

Contributions welcome — code, docs, issue triage, bug reports. See CONTRIBUTING.md.

Known Limitations

  1. Loop on non-code tasks (research/docs) fails silently — workers produce no diff, loop reports failure
  2. Workers inherit parent env — project-specific env vars leak into worker shells; sanitize before overnight runs
  3. Context budget is per-session — multi-day runs may exhaust context; use /handoff + /pickup

License

MIT

Reviews (0)

No results found