codemind-okf 1.0.2

CodeMind OKF CLI — Generate AI-ready knowledge bundles for any codebase. Works with Cursor, Antigravity, GitHub Copilot, and any MCP-compatible AI IDE.

CodeMind OKF CLI

Generate AI-ready knowledge bundles from any codebase. Works with Cursor, Antigravity, GitHub Copilot, and any MCP-compatible AI IDE.

Install

pip install codemind-okf

Quick Start

# 1. Navigate to your project
cd my-project

# 2. Generate the OKF knowledge bundle (zero LLM cost — pure AST)
codemind index .

# 3. Drop AI IDE config files so Cursor/Antigravity/Copilot use the bundle
codemind init

# 4. Open your IDE — AI now reads .okf/index.md as its primary context!

Commands

codemind index <path>

Crawls and indexes a project into an .okf/ knowledge bundle.

codemind index .                         # Index current directory
codemind index /path/to/project          # Index a specific directory
codemind index . --lang python           # Only index Python files
codemind index . --overwrite             # Re-index from scratch

codemind init

Drops AI IDE instruction files that tell Cursor, Antigravity, and Copilot to use .okf/ as context.

codemind init                            # Create all AI IDE config files
codemind init --no-copilot              # Skip GitHub Copilot instructions

Files created:

File	AI Tool
.cursorrules	Cursor AI
.agents/AGENTS.md	Antigravity IDE & compatible agents
.github/copilot-instructions.md	GitHub Copilot

codemind status

Show the current bundle statistics.

codemind status
codemind status /path/to/project

codemind watch

Watches the project directory for real-time file changes and updates .okf/ incrementally in milliseconds.

codemind watch                          # Watch current directory

codemind audit

Performs an architectural health and code density audit of your codebase using your OKF bundle.

codemind audit                          # Audit current project codebase

codemind mcp

Launches the Model Context Protocol (MCP) server over STDIO.

Connects CodeMind natively to Cursor, Claude Desktop, Antigravity, or Zed.

Add to Cursor / Claude Desktop (mcp.json):

{
  "mcpServers": {
    "codemind": {
      "command": "codemind",
      "args": ["mcp"]
    }
  }
}

Once configured, Cursor or Claude Desktop will automatically call CodeMind tools (search_bundle, read_module, get_project_index, trace_dependencies) whenever you ask questions about your codebase!

How It Works

Your Codebase  →  AST Parser  →  ModuleSummary  →  .okf/modules/*.md
                                                    .okf/index.md
                                                    .okf/log.md
                                                           ↓
                              AI IDE reads .cursorrules → reads .okf/index.md
                              → understands full architecture instantly

All parsing is deterministic and free — no LLM calls, no API keys required.

Output Structure

.okf/
├── index.md              ← Master architecture map (AI IDE reads this first)
├── log.md                ← Generation audit trail
└── modules/
    ├── src-auth-router.md
    ├── src-database-models.md
    └── src-api-endpoints.md

Each module file contains structured YAML frontmatter:

---
type: api
title: Auth Router
description: Handles user authentication endpoints including login, logout, and token refresh.
resource: src/auth/auth_router.py
tags:
  - api
  - auth
  - fastapi
key_functions:
  - login
  - logout
  - refresh_token
timestamp: 2025-07-23
---

Supported Languages

• Python (.py) — Full AST analysis
• JavaScript (.js, .jsx) — AST + regex parsing
• TypeScript (.ts, .tsx) — AST + regex parsing

License

MIT © CodeMind