openclaw-claude-codeProgrammatically control Claude Code via MCP protocol to execute commands, manage sessions, search and edit code, and coordinate agent teams for development...
Install via ClawdBot CLI:
clawdbot install Enderfga/openclaw-claude-codeControl Claude Code via MCP (Model Context Protocol). This skill unleashes the full power of Claude Code for openclaw agents, including persistent sessions, agent teams, and advanced tool control.
# Start a persistent Claude session for your project
claude-code-skill session-start myproject -d ~/project \
--permission-mode plan \
--allowed-tools "Bash,Read,Edit,Write,Glob,Grep" \
--max-budget 2.00
# Send a complex task (Claude will autonomously use tools)
claude-code-skill session-send myproject "Find all TODO comments and create GitHub issues" --stream
# Check progress
claude-code-skill session-status myproject# Connect to Claude Code MCP
claude-code-skill connect
claude-code-skill status
claude-code-skill tools
# Direct tool calls (no persistent session)
claude-code-skill bash "npm test"
claude-code-skill read /path/to/file.ts
claude-code-skill glob "**/*.ts" -p ~/project
claude-code-skill grep "TODO" -p ~/project -c
claude-code-skill call Write -a '{"file_path":"/tmp/test.txt","content":"Hello"}'
# Disconnect
claude-code-skill disconnect# Basic start
claude-code-skill session-start myproject -d ~/project
# With custom API endpoint (for Gemini/GPT proxy)
claude-code-skill session-start gemini-task -d ~/project \
--base-url http://127.0.0.1:8082 \
--model gemini-2.0-flash
# With permission mode (plan = preview changes before applying)
claude-code-skill session-start review -d ~/project --permission-mode plan
# With tool whitelist (auto-approve these tools)
claude-code-skill session-start safe -d ~/project \
--allowed-tools "Bash(git:*),Read,Glob,Grep"
# With budget limit
claude-code-skill session-start limited -d ~/project --max-budget 1.50
# Full configuration
claude-code-skill session-start advanced -d ~/project \
--permission-mode acceptEdits \
--allowed-tools "Bash,Read,Edit,Write" \
--disallowed-tools "Task" \
--max-budget 5.00 \
--model claude-opus-4-5 \
--append-system-prompt "Always write tests" \
--add-dir "/tmp,/var/log"Permission Modes:
| Mode | Description |
|------|-------------|
| acceptEdits | Auto-accept file edits (default) |
| plan | Preview changes before applying |
| default | Ask for each operation |
| bypassPermissions | Skip all prompts (dangerous!) |
| delegate | Delegate decisions to parent |
| dontAsk | Never ask, reject by default |
# Basic send (blocks until complete)
claude-code-skill session-send myproject "Write unit tests for auth.ts"
# Streaming (see progress in real-time)
claude-code-skill session-send myproject "Refactor this module" --stream
# With custom timeout
claude-code-skill session-send myproject "Run all tests" -t 300000# List active sessions
claude-code-skill session-list
# Get detailed status
claude-code-skill session-status myproject
# View conversation history
claude-code-skill session-history myproject -n 50
# Pause and resume
claude-code-skill session-pause myproject
claude-code-skill session-resume-paused myproject
# Fork a session (create a branch for experiments)
claude-code-skill session-fork myproject myproject-experiment
# Stop
claude-code-skill session-stop myproject
# Restart a failed session
claude-code-skill session-restart myproject# Browse all Claude Code sessions
claude-code-skill sessions -n 20
# Search sessions by project
claude-code-skill session-search --project ~/myapp
# Search by time
claude-code-skill session-search --since "2h"
claude-code-skill session-search --since "2024-02-01"
# Search by query
claude-code-skill session-search "bug fix"
# Resume a historical session
claude-code-skill resume <session-id> "Continue where we left off" -d ~/project# Read multiple files at once
claude-code-skill batch-read "src/**/*.ts" "tests/**/*.test.ts" -p ~/projectDeploy multiple Claude agents working together on complex tasks.
# Define a team of agents
claude-code-skill session-start team-project -d ~/project \
--agents '{
"architect": {
"description": "Designs system architecture",
"prompt": "You are a senior software architect. Design scalable, maintainable systems."
},
"developer": {
"description": "Implements features",
"prompt": "You are a full-stack developer. Write clean, tested code."
},
"reviewer": {
"description": "Reviews code quality",
"prompt": "You are a code reviewer. Check for bugs, style issues, and improvements."
}
}' \
--agent architect
# Switch between agents mid-conversation
claude-code-skill session-send team-project "Design the authentication system"
# (architect responds)
claude-code-skill session-send team-project "@developer implement the design"
# (developer agent takes over)
claude-code-skill session-send team-project "@reviewer review the implementation"
# (reviewer agent takes over)# Code review team
claude-code-skill session-start review -d ~/project \
--agents '{
"security": {"prompt": "Focus on security vulnerabilities"},
"performance": {"prompt": "Focus on performance issues"},
"quality": {"prompt": "Focus on code quality and maintainability"}
}' \
--agent security
# Full-stack team
claude-code-skill session-start fullstack -d ~/project \
--agents '{
"frontend": {"prompt": "React/TypeScript frontend specialist"},
"backend": {"prompt": "Node.js/Express backend specialist"},
"database": {"prompt": "PostgreSQL/Redis database specialist"}
}' \
--agent frontend# Allow specific tools with patterns
--allowed-tools "Bash(git:*,npm:*),Read,Edit"
# Deny dangerous operations
--disallowed-tools "Bash(rm:*,sudo:*),Write(/etc/*)"
# Limit to specific tool set
--tools "Read,Glob,Grep"
# Disable all tools
--tools ""# Replace system prompt completely
--system-prompt "You are a Python expert. Always use type hints."
# Append to existing prompt
--append-system-prompt "Always run tests after changes."# Resume with fork (create a branch)
--resume <session-id> --fork-session
# Use custom UUID for session
--session-id "550e8400-e29b-41d4-a716-446655440000"
# Add additional working directories
--add-dir "/var/log,/tmp/workspace"Use --base-url to route requests through a proxy, enabling other models (Gemini, GPT) to power Claude Code:
# Use Gemini via claude-code-proxy
claude-code-skill session-start gemini-task -d ~/project \
--base-url http://127.0.0.1:8082 \
--model claude-3-5-sonnet-20241022 # Proxy will map to Gemini
# Use GPT via proxy
claude-code-skill session-start gpt-task -d ~/project \
--base-url http://127.0.0.1:8082 \
--model claude-3-haiku-20240307 # Proxy will map to GPTNote: Requires claude-code-proxy running on port 8082 with proper API keys configured.
# Start the proxy
cd ~/clawd/claude-code-proxy && source .venv/bin/activate
uvicorn server:app --host 127.0.0.1 --port 8082
# ❌ Bad: Multiple disconnect/reconnect cycles
claude-code-skill bash "step1"
claude-code-skill bash "step2"
# ✅ Good: Single persistent session
claude-code-skill session-start task -d ~/project
claude-code-skill session-send task "Do step1 then step2"--stream for long-running tasks
claude-code-skill session-send task "Run full test suite" --stream
--max-budget 2.00 # Stop after $2 of API usage
--permission-mode plan # Preview before applying
claude-code-skill session-fork main experimental
claude-code-skill session-send experimental "Try risky refactor"# If session fails:
claude-code-skill session-status myproject # Check what happened
claude-code-skill session-history myproject -n 20 # See recent events
claude-code-skill session-restart myproject # Restart from last good state
# If you need to start over:
claude-code-skill session-stop myproject
claude-code-skill session-start myproject -d ~/project --resume <old-session-id>openclaw agent
↓
claude-code-skill CLI (this tool)
↓ HTTP
sasha-doctor API (:18795)
↓ MCP
claude mcp serve (Claude Code)
↓
Your files & toolsAll Claude Code tools are accessible:
| Tool | Description |
|------|-------------|
| Bash | Execute shell commands |
| Read | Read file contents |
| Write | Create/overwrite files |
| Edit | Edit files with string replacement |
| Glob | Find files by pattern |
| Grep | Search file contents |
| Task | Launch sub-agents |
| WebFetch | Fetch web content |
| WebSearch | Search the web |
| Git* | Git operations |
| AskUserQuestion | Interactive prompts |
| ... | and 10+ more |
claude-code-skill session-start review -d ~/myapp \
--permission-mode plan \
--agents '{"security":{"prompt":"Focus on security"},"quality":{"prompt":"Focus on quality"}}' \
--agent security
claude-code-skill session-send review \
"Review all TypeScript files in src/, check for security issues and code quality problems" \
--streamclaude-code-skill session-start test -d ~/myapp \
--allowed-tools "Bash(npm:*,git:*),Read,Write" \
--max-budget 1.00
claude-code-skill session-send test \
"Find all untested functions, write unit tests, run tests, fix failures"claude-code-skill session-start debug -d ~/myapp \
--agents '{
"detective": {"prompt": "Find the root cause of bugs"},
"fixer": {"prompt": "Implement fixes"},
"tester": {"prompt": "Verify fixes work"}
}' \
--agent detective
claude-code-skill session-send debug "We have a memory leak in the API server" --stream
# Detective investigates, then hands off to fixer, then to testerWhen openclaw needs to perform complex coding tasks:
# From within openclaw agent context:
openclaw skills run claude-code-skill -- session-start task -d ~/project
openclaw skills run claude-code-skill -- session-send task "Implement feature X" --stream
openclaw skills run claude-code-skill -- session-status taskOr use the skill programmatically via sasha-doctor HTTP API (see TOOLS.md section 3).
qmd tool)Generated Mar 1, 2026
A software development team uses persistent sessions to iteratively refactor legacy codebases. Claude autonomously analyzes code, identifies technical debt, and applies safe refactorings with plan permission mode to preview changes before committing, ensuring quality and reducing manual review time.
DevOps engineers integrate the skill into CI/CD pipelines to automate testing and deployment tasks. Using direct MCP tools like bash and grep, they run tests, check logs for errors, and update configurations programmatically, speeding up release cycles and improving reliability.
A tech company employs agent teams with architect, developer, and reviewer roles to conduct thorough code reviews. Each agent specializes in different aspects, allowing collaborative analysis of pull requests, catching bugs early, and enforcing coding standards without human intervention.
Documentation teams use batch operations to read source files and generate or update documentation. Claude searches for comments and code patterns, then writes formatted docs, keeping documentation synchronized with code changes and reducing manual effort.
Security analysts leverage the skill to scan codebases for vulnerabilities. Using grep and glob tools, Claude identifies risky patterns like hardcoded secrets or outdated dependencies, then creates reports or issues, enhancing security posture through continuous monitoring.
Offer the skill as a cloud-based service with tiered subscriptions. Provide features like advanced session management, team collaboration tools, and analytics dashboards, targeting enterprises that need scalable AI-powered development automation.
Provide professional services to customize and integrate the skill into clients' existing workflows. Offer training, support, and tailored configurations for specific industries, generating revenue through project-based contracts and ongoing maintenance fees.
Distribute the core skill as open source to build a community and drive adoption. Monetize through premium add-ons like enhanced agent teams, enterprise security features, or priority support, leveraging a large user base for upselling.
💬 Integration Tip
Start with direct MCP tools for simple tasks like file operations, then gradually adopt persistent sessions for complex workflows, ensuring proper permission modes to maintain control and safety.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Provides a 7-step debugging protocol plus language-specific commands to systematically identify, verify, and fix software bugs across multiple environments.
A comprehensive skill for using the Cursor CLI agent for various software engineering tasks (updated for 2026 features, includes tmux automation guide).
Write, run, and manage unit, integration, and E2E tests across TypeScript, Python, and Swift using recommended frameworks.
Control and operate Opencode via slash commands. Use this skill to manage sessions, select models, switch agents (plan/build), and coordinate coding through Opencode.
Coding style memory that adapts to your preferences, conventions, and patterns for consistent coding.