claude-delegator  by jarrodwatts

AI-powered delegation for enhanced code analysis and implementation

Created 1 month ago
835 stars

Top 42.4% on SourcePulse

GitHubView on GitHub
1 Expert Loves This Project
Project Summary

Delegates complex tasks to specialized GPT experts within Claude Code, enhancing AI-assisted development. It targets users of Claude Code seeking advanced capabilities for architecture, security, and code review, offering a team of AI specialists to analyze and implement solutions, thereby improving efficiency and code quality.

How It Works

This project leverages Claude's native MCP (Meta-Cognitive Programming) to route user requests to five distinct GPT expert subagents: Architect, Plan Reviewer, Scope Analyst, Code Reviewer, and Security Analyst. Each expert operates in either 'Advisory' (read-only analysis) or 'Implementation' (workspace-write) modes, automatically selected based on the task. Claude synthesizes the GPT expert's output, providing a coherent, judgment-applied response rather than raw passthrough. This approach offers specialized AI assistance tailored to specific development challenges.

Quick Start & Requirements

  • Installation:
    1. marketplace add jarrodwatts/claude-delegator
    2. plugin install claude-delegator
    3. /claude-delegator:setup
  • Prerequisites:
    • Codex CLI: npm install -g @openai/codex
    • Codex Authentication: codex login
    • Requires a Claude Code instance.
  • Setup: The /claude-delegator:setup command configures the MCP server. Manual configuration details are available in the README.

Highlighted Details

  • Five Domain Experts: Dedicated specialists for Architecture, Plan Review, Scope Analysis, Code Review, and Security.
  • Dual Operating Modes: Experts support 'Advisory' (analysis) and 'Implementation' (changes) modes, auto-selected by Claude.
  • Automatic Task Routing: Claude intelligently detects when to delegate tasks to the appropriate expert based on the request.
  • Synthesized Responses: Claude interprets and integrates expert feedback, delivering a refined output.
  • Customizable Prompts: Expert system prompts are located in the prompts/ directory for user customization.

Maintenance & Community

The README mentions expert prompts adapted from 'oh-my-opencode' by @code-yeongyu. No specific details regarding active contributors, sponsorships, or community channels (e.g., Discord, Slack) are provided.

Licensing & Compatibility

  • License: MIT License.
  • Compatibility: The MIT license is permissive, generally allowing for commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

The tool is not intended for simple file operations, initial attempts at fixes, or trivial questions. If automatic expert triggering fails, users may need to use more explicit prompts. Troubleshooting guidance is provided for common issues such as MCP server discovery or authentication problems.

Health Check
Last Commit

1 month ago

Responsiveness

Inactive

Pull Requests (30d)
2
Issues (30d)
2
Star History
90 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.