What is Aidemd?
Aidemd is an MCP server that brings intent-driven development to any AI-powered IDE by managing .aide spec files that live next to your code. It provides agents with tools to discover, scaffold, read, validate, and inspect these intent, research, plan, and QA specs, enabling architects, implementors, and QA to work from a shared domain context.
How to use Aidemd?
Run npx @aidemd-mcp/server@latest init in your project, then execute /aide:init inside Claude Code to bootstrap the methodology. Alternatively, add the server manually to your MCP client configuration (e.g., .mcp.json, claude_desktop_config.json, or client-specific MCP files) with the command npx -y @aidemd-mcp/server@latest. Once connected, agents can invoke tools such as aide_discover, aide_scaffold, aide_read, aide_validate, aide_inspect, aide_init, and aide_upgrade.
Key features of Aidemd
- Project-wide
.aide spec discovery with a progressive disclosure tree
- One-command project bootstrap via
aide_init
- Automatic naming convention enforcement through
aide_scaffold
- Health-check validation detecting orphaned specs and naming conflicts
- Code introspection returning JSDoc, signatures, and kind for named symbols
- Upgrade drift detection comparing artifacts against canonical versions
Use cases of Aidemd
- Bootstrapping the full AIDE methodology into a new or existing project
- Scaffolding intent and research specs with correct naming conventions in the right directory
- Validating the health of all
.aide files across the codebase to prevent drift
- Inspecting a symbol's contract (JSDoc, signature, kind) without opening its source file
- Detecting and applying upgrades to methodology artifacts as canonical versions change
コメント