AIM 🎯 — one context, every AI assistant¶

AIM (AI Memory / Mind) is a zero-dependency command-line engine that keeps your project's memory, tasks, docs, and rules in one place and compiles them into the native instruction files every AI coding assistant expects — so Claude Code, Cursor, Windsurf, GitHub Copilot, Gemini, and Antigravity all share the same brain.

It also serves that context live over the Model Context Protocol (MCP), detects context drift before it makes agents confidently wrong, and syncs your tasks to GitHub Issues & Projects.

One source of truth

Write rules once in .ai-context/. aim sync compiles CLAUDE.md, AGENTS.md, .cursorrules, .windsurfrules, GEMINI.md, and copilot-instructions.md — no more copy-paste drift across tools.
Memory that ranks itself

Persist decisions and corrections; recall the most relevant ones with get_memory_context (relevance × importance × recency) as working memory for any agent.
Spec-driven tasks

aim spec new scaffolds requirements / design / tasks with EARS acceptance criteria, dependency-ordered and ready to decompose.
Context health (aim doctor)

Deterministically flags stale memory (vs git history), broken references, and duplicate IDs — CI-friendly, no LLM required.
Zero-dependency MCP server

aim mcp exposes tasks, docs, and memory to any MCP client over stdio — pure stdlib, nothing extra to install.
Local web dashboard

aim browser opens a localhost Kanban board, knowledge graph, and a Health tab — token-protected, never exposed to the network.

🚀 Quick start¶

# Install
pip install git+https://github.com/phuonghx/aim-cli.git

# Initialize in your project, then compile instructions for every AI tool
aim init
aim sync

# Connect it live to Claude Code (or any MCP client)
claude mcp add aim -- aim mcp

Already have scattered rules? Pull existing CLAUDE.md / .cursorrules / AGENTS.md into AIM in one step:

aim ingest --dry-run     # preview
aim ingest && aim sync   # consolidate, then re-emit everywhere

📚 Explore the docs¶

CLI Reference — every command and flag.
AI Agent Integration — how each assistant consumes AIM.
Web Dashboard — the visual control hub.
Templates — code generation with variables and case helpers.
Demo Walkthrough — a ready-made example workspace.
Roadmap — where AIM is headed.

Why AIM?¶

Modern teams use several AI assistants at once, and each wants its rules in a different file. Without a single source of truth those files drift, agents act on stale assumptions, and the same mistakes repeat across sessions and tools. AIM fixes the context layer: one place to write it, every tool to read it, and a doctor to keep it honest.

Free and open source under the MIT License.