devplan

DevPlan MCP Server exposes a structured workflow to generate, enhance, validate, and execute Haiku-executable development plans for coding tasks. The server implements a scaffold → enhance → validate loop, pairing with Claude Code tooling to interview you, produce a PROJECT_BRIEF, scaffold initial code, and iteratively refine it until it passes validation. Tools and concepts you’ll leverage include plan generation, code enhancement by Claude models, validation checks that ensure completeness and correct imports, and a learning loop that captures issues as lessons for future projects. When you invoke the server via its MCP interface, you can generate plans, create executors and verifiers for multiple targets, and monitor progress through task tooling integrated into Claude Code.

Prerequisites:

• Node.js and npm (for npx-based setup) or a compatible runtime as required by the chosen transport
• Access to the Claude Code MCP ecosystem (as described in the DevPlan README)

Install steps:

1. Ensure you have npx available (comes with npm):
   • Check: node -v && npm -v
2. Add the DevPlan MCP server vianpx (this pulls the server tooling and dependencies):
   • npx -y mmorris35/ClaudeCode-DevPlanBuilder
3. Configure your environment to point to the MCP server transport as described in the README (e.g., SSE transport URL for Cloudflare Workers or other supported transports).
4. Optionally register the server in your Claude configuration (depending on your setup) by adding it under mcpServers with the appropriate transport details.

Post-install:

• Verify that the MCP server can be reached by querying its status or starting a small plan-generation task as described in install/setup docs for DevPlan.

Tips and notes:

• If you encounter transport/URL issues, verify your SSE endpoint or Cloudflare Workers deployment is accessible from your network.
• The DevPlan workflow emphasizes a strict Haiku-executable standard; ensure all code blocks are complete with imports and no placeholders before validation.
• When generating executors/verifiers for different targets, use the appropriate adapter (Claude, Cursor, Aider, Cline, Windsurf, or Generic) to produce the correct files in your project.
• Common issues often involve missing imports or incomplete code blocks in generated artifacts; rerun the enhance/validate steps after addressing any flagged items.
• Environment variables can be used to customize transport endpoints, authentication, or feature flags; document any required vars in your local environment or CI configuration.