Claude Code Integration

Claude Code integration for OpenClaw. This skill provides interfaces to: - Query Claude Code documentation from https://code.claude.com/docs - Manage subagents and coding tasks - Execute AI-assisted coding workflows - Access best practices and common workflows Use this skill when users want to: - Get help with coding tasks - Query Claude Code documentation - Manage AI-assisted development workflows - Execute complex programming tasks

Security Scan

The skill is internally consistent with a documentation-and-wrapper utility that uses OpenClaw's native capabilities; there are no obvious requests for unrelated credentials, high-risk installers, or hidden exfiltration in the provided files.

Purpose & Capability

The name/description claim a Claude Code integration that surfaces documentation and coordinates tasks via OpenClaw. The package includes a local CLI-like Python script and an install script that copies files into the OpenClaw skills directory. Requiring OpenClaw (checked in install.sh) matches the stated purpose. The only mild inconsistency: the documentation text mentions ANTHROPIC_API_KEY and remote Claude installers, but the skill itself declares no required environment variables because it intends to leverage OpenClaw for runtime access.

Instruction Scope

SKILL.md and README are focused on documentation queries and subagent/task workflows routed through OpenClaw. The instructions explicitly indicate that task execution happens through OpenClaw's subagent system and reference OpenClaw exec/read-write tools rather than direct network calls from the skill. The documentation includes example curl/powershell install snippets for external Claude binaries, but those are shown in doc text and not invoked by the skill itself.

Install Mechanism

No remote downloads are performed by install.sh; it requires a local openclaw binary and copies repository files into $HOME/.openclaw/skills/claude-code and makes the Python script executable. That is a low-risk, standard local install. The README contains examples that pipe remote installers from third-party domains (curl | bash) but those are only examples in documentation, not used by the skill's install.sh.

Credentials

The skill declares no required environment variables (consistent with relying on OpenClaw), which is proportional. Documentation content references ANTHROPIC_API_KEY and other Claude/Anthropic config locations — that's informative for users who separately install/authorize Claude Code, but it is not required by this skill. This mismatch is acceptable but should be noted: if the skill were later extended to call Claude/Anthropic directly, an API key would be expected.

Persistence & Privilege

always:false and no modifications to other skills or system-wide configuration are present. install.sh writes only to the skill's directory under ~/.openclaw/skills, which is normal for a user-level skill install.

Assessment

This package appears to be a documentation/wrapper skill that delegates real execution to OpenClaw. Before installing: (1) Inspect claude-code.py for any network calls or subprocess.exec that might contact external services (we reviewed the provided docs and install script and saw no active remote downloads, but you should confirm the script body does not perform HTTP requests). (2) Do not run arbitrary curl | bash commands shown in the documentation unless you trust the remote host — the docs include example installer commands for external Claude binaries which are separate from this skill. (3) If you expect the skill to create/manage subagents that interact with Claude/Anthropic directly, verify whether it will require or accept an API key and where that key would be stored. (4) Install via the included install.sh only after confirming the script and Python code match your expectations. Overall the skill looks coherent and proportionate, but standard caution about reviewing code and not blindly running remote installers applies.

This skill integrates Claude Code capabilities into OpenClaw, providing access to AI-assisted coding workflows, documentation, and best practices.

What You Can Do

📚 Documentation Queries

• Query Claude Code documentation
• Get best practices and workflows
• Learn about settings and customization
• Troubleshoot common issues

🤖 Subagent Management

• Create coding subagents
• Manage agent teams
• Execute complex development tasks
• Automate code reviews and PR workflows

🛠️ Development Workflows

• Best practices for AI-assisted coding
• Common workflows and patterns
• Settings and configuration
• Troubleshooting guidance

Usage Examples

Query Documentation

# Get documentation about a specific topic
claude-code query "subagents"
claude-code query "best practices"
claude-code query "settings"

Execute Coding Task

# Create a coding subagent for a complex task
claude-code task --description "Fix the login bug" --priority high
claude-code task --description "Refactor the database layer" --model claude-3-5-sonnet

List Available Commands

# Show all available commands
claude-code --help

Available Commands

query

Query Claude Code documentation for a specific topic.

Usage:

claude-code query <topic>

Examples:

claude-code query "subagents"
claude-code query "agent-teams"
claude-code query "best practices"
claude-code query "common workflows"
claude-code query "settings"
claude-code query "troubleshooting"

Topics include:

• Subagents and agent teams
• Best practices and workflows
• Settings and customization
• Troubleshooting guide
• Plugins and extensions
• MCP (Model Context Protocol)
• Headless/Programmatic usage

task

Create and execute a coding subagent task.

Usage:

claude-code task --description "<task description>" [--priority <level>] [--model <model-name>]

Options:

• --description, -d: Task description (required)
• --priority, -p: Task priority (low/medium/high, default: medium)
• --model, -m: Model to use (optional, uses default if not specified)

Examples:

claude-code task --description "Implement user authentication module"
claude-code task --description "Refactor database queries" --priority high
claude-code task --description "Write unit tests for the API" --model claude-3-5-sonnet

docs

Get overview of Claude Code documentation sections.

Usage:

claude-code docs [section]

Sections:

• quickstart - Getting started guide
• best-practices - AI coding best practices
• common-workflows - Typical development workflows
• settings - Customization options
• troubleshooting - Common issues and solutions
• all - Full documentation overview (default)

Examples:

claude-code docs
claude-code docs quickstart
claude-code docs best-practices
claude-code docs troubleshooting

info

Display Claude Code configuration and status.

Usage:

claude-code info

Output includes:

• Version information
• Available subagents
• Configured models
• MCP servers status

Integration with OpenClaw

This skill works seamlessly with OpenClaw's native capabilities:

• Subagents: Claude Code subagents complement OpenClaw's subagent system
• Code Execution: Use with OpenClaw's exec tool for complete development workflow
• File Management: Combine with OpenClaw's read/write tools for full codebase management
• Sessions: Claude Code tasks integrate with OpenClaw's session management

Example Workflows

Complex Bug Fix

# 1. Query best practices for debugging
claude-code query "debugging best practices"

# 2. Create a subagent to investigate and fix
claude-code task --description "Find and fix the null pointer exception in userService.js" --priority high

# 3. Review the changes
claude-code query "code review best practices"

New Feature Development

# 1. Get best practices for the feature type
claude-code query "API design best practices"

# 2. Create development task
claude-code task --description "Implement REST API for user management" --priority medium

# 3. Check settings for code style
claude-code query "code style settings"

Code Review Automation

# 1. Query PR review best practices
claude-code query "PR review workflows"

# 2. Set up automated review task
claude-code task --description "Review all PRs in the last week" --priority low

Configuration

Environment Variables

Not required for basic usage. Claude Code integration uses OpenClaw's native capabilities.

Models

Uses OpenClaw's configured default models. Override per task with --model option.

Subagent Limits

Managed by OpenClaw's subagent configuration (default: 8 concurrent subagents).

Notes

• This skill provides a wrapper around Claude Code documentation and workflows
• Complex coding tasks are executed through OpenClaw's native subagent system
• For direct Claude Code CLI usage, install Claude Code separately from https://claude.com/code
• All task execution happens through OpenClaw's secure agent infrastructure

See Also

• Claude Code Official Docs: https://code.claude.com/docs
• OpenClaw Subagents: Use OpenClaw's native subagent functionality
• Best Practices: Integrated from Claude Code guidelines