aida-mcp 0.7.0

AIDA MCP server - Developer actions (lint, test, typecheck) for monorepos

AIDA MCP

AIDA gives your AI coding assistant repository-specific rules, validation, and Git workflow guardrails — across any MCP-compatible tool.

Instead of maintaining separate prompt files for each assistant, you configure behavior once in .aida/ and every tool picks it up.

What AIDA Does for You

As a developer, AIDA works in the background. Your AI assistant automatically:

• Receives project-specific coding rules and conventions when you start a task
• Runs the right linters, formatters, and tests for the files you changed — before you commit
• Creates commits and PRs that follow your team's message format, required trailers, and templates

You don't need to learn new commands. AIDA teaches your assistant your team's workflow.

As a repo maintainer, AIDA lets you:

• Define validation pipelines that run automatically on changed files
• Enforce commit message format, risk labels, ticket references
• Write project-specific rules your assistant will follow
• Configure all of this in version-controlled YAML files under .aida/

Install

uv tool install aida-mcp    # recommended (uv is a fast Python package manager: brew install uv)
pip install aida-mcp         # alternative

Quick Start

If your repo already has AIDA configured

Just install aida-mcp and reload your AI tool. The .mcp.json / .cursor/mcp.json in the repo tells your tool to connect to AIDA. Everything works automatically.

If you're setting up AIDA in a new repo

cd /path/to/your/repo
aida-mcp init

This creates .aida/ config files, MCP wiring, and rule files. Commit the result and reload your AI tool.

For the full setup guide, see Configuring a Repository.

Supported Tools

Cursor, Claude Code, OpenAI Codex, JetBrains AI Chat, and Junie. Any MCP-compatible tool can be added.

Customizing Behavior

Preferences

Control how AIDA behaves per-developer via .aida/preferences.local.yaml (gitignored):

preferences:
  validation:
    pre_commit: ask      # auto | ask | disabled
  commit:
    mode: ask            # auto | ask | disabled

Team defaults live in .aida/preferences.yaml (checked in).

Native Mode (No MCP Server)

If you prefer not to run the MCP server, export rules as static files your AI tool reads directly:

aida-mcp rules export

This writes self-contained rule files with your project's conventions, validation commands, and commit templates. Re-run after config changes; aida-mcp doctor warns when stale.

CLI Commands

Command	Purpose
aida-mcp init	Bootstrap AIDA in a repository
aida-mcp doctor	Check repository health
aida-mcp migrate	Upgrade from older AIDA layout
aida-mcp validate --scope pre_commit	Run validation on changed files
aida-mcp commit --type feat --title "..."	Policy-compliant commit
aida-mcp rules export	Export static native rule files
aida-mcp install-mcp --target claude	Install/refresh MCP wiring

Full CLI reference: cli-reference.md

Documentation

• For Developers — how AIDA works in your daily workflow
• Configuring a Repository — setup guide for repo maintainers
• CLI Reference — all commands and flags
• Rules Retrieval — how get_rules scoring and selection works
• Validation Tools — planner + executor architecture
• External Validation Interface — wiring custom scripts

Requirements

• Python ≥3.12, <3.15
• An MCP-compatible AI tool
• A repository you want to configure