This repository provides a configuration framework for Claude Code, enhancing its capabilities with Model Context Protocol (MCP) servers, custom commands, and automated workflows. It targets Claude Code users seeking to automate complex development tasks, offering specialized agents for code investigation, flow mapping, planning, and review, thereby streamlining problem-solving and code quality assurance.

How It Works

The setup integrates MCP servers like Context7 and Sequential Thinking for advanced reasoning and context retrieval. It introduces custom commands (/task_hard, /code-review, /commit) that orchestrate multi-agent workflows. A hook system automates directory creation and task initiation. Key agents include an investigator that uses sequential thinking and keyword prioritization for codebase analysis, a code-flow-mapper for tracing execution paths, a planner that synthesizes analysis into actionable plans, and a code-reviewer for quality checks.

Quick Start & Requirements

• Installation: Install uv (pip install uv), clone the repository (git clone claude-setup), and navigate into the directory (cd claude-setup).
• Prerequisites: Claude Code installation, Python 3.8+, uv package manager, and Node.js (for MCP server execution via npx).
• Configuration: Copy .claude/ and .mcp.json to the project root and ensure hook script permissions (chmod +x .claude/hooks/task_hard_prep_hook.py).
• Resources: No external links provided; the README serves as the primary documentation.

Highlighted Details

• Custom Commands: /task_hard orchestrates a multi-agent workflow for complex problem-solving, including automated directory creation, keyword extraction, codebase investigation (generating INVESTIGATION_REPORT.md), code flow mapping (FLOW_REPORT.md), and planning (PLAN.md). /code-review analyzes uncommitted changes for quality and security. /commit facilitates conventional commits.
• Specialized Agents: investigator uses sequential thinking and prioritizes files based on extracted keywords. code-flow-mapper traces execution paths and file interdependencies. planner synthesizes reports into actionable plans. code-reviewer provides prioritized feedback on code quality.
• Hook System: The UserPromptSubmit hook automates the creation of task-specific directories within claude-code-storage/.
• MCP Servers: Context7 provides library documentation and code context; Sequential Thinking enables advanced reasoning capabilities.

Maintenance & Community

The provided README does not detail specific contributors, community channels (e.g., Discord, Slack), or a public roadmap. The "Contributing" section outlines a standard pull request workflow for contributions.

Licensing & Compatibility

The repository's license is not specified in the README, which is a critical omission for evaluating adoption and commercial use.

Limitations & Caveats

Hooks relying on relative paths may fail if Claude Code is executed from a subdirectory within a monorepo. While $CLAUDE_PROJECT_DIR is suggested as a workaround, its cross-platform compatibility is noted as potentially problematic. The absence of a specified license prevents a full assessment of compatibility and usage restrictions.