DevMind — Semantic Codebase Memory & Intelligence Engine

"Your codebase finally has a persistent memory."
An open-source, local-first Codebase Memory and Static Intelligence Engine powered by Cognee, FastEmbed, LanceDB, and FastMCP.

⚡ Quickstart (30 Seconds)

Get up and running in any project with just three commands:

# 1. Install via PyPI
pip install devmind-cli

# 2. Interactive Setup (Configure free Cloud API or 100% Local Ollama)
devmind init

# 3. Index Codebase & Ask Questions
devmind remember
devmind ask "Where is the authentication flow handled and how are tokens validated?"

🦙 100% Local & Offline Setup with Ollama (Step-by-Step)

If you want complete data privacy with zero tokens leaving your machine, you can run DevMind entirely on local models using Ollama:

Step 1: Install Ollama

Download and install Ollama for your operating system:

• macOS / Windows: Download from ollama.com
• Linux:

  curl -fsSL https://ollama.com/install.sh | sh
  

Step 2: Download a Coding Model

Open your terminal and pull your preferred model:

# Recommended lightweight model (Fast & works on any laptop):
ollama pull llama3.2

# Recommended for high-accuracy coding tasks:
ollama pull qwen2.5-coder:7b
# or
ollama pull mistral

Step 3: Ensure Ollama Server is Running

Ollama runs in the background automatically. To verify, run:

ollama serve

(By default, Ollama is accessible at http://localhost:11434)

Step 4: Configure DevMind for Ollama

Run the DevMind setup wizard:

devmind init

1. Select option [5] 🦙 Ollama.
2. Press Enter to accept the default base URL (http://localhost:11434).
3. Enter the model name you pulled (e.g. llama3.2 or qwen2.5-coder:7b).
4. DevMind will test the local connection and save your configuration globally.

💡 Zero-Cost Local Embeddings: DevMind uses local CPU-accelerated FastEmbed (BAAI/bge-small-en-v1.5) for vector indexing. No embeddings are sent to external APIs!

⚡ Free Cloud AI Providers (Groq & Google Gemini)

If you prefer cloud models without needing local GPU/RAM:

• ⚡ Groq (Recommended - Ultra Fast & 100% Free):
  1. Get a free API key at console.groq.com/keys.
  2. Run devmind init and choose [1] Groq.
• ♊ Google Gemini (Generous Free Tier):
  1. Get a free API key at aistudio.google.com/app/apikey.
  2. Run devmind init and choose [2] Gemini.
• 🟣 Anthropic Claude / 🟢 OpenAI / 🌐 OpenRouter: Run devmind init to configure Claude (claude-3-5-sonnet), OpenAI (gpt-4o-mini, o3-mini), or OpenRouter keys.

🛠️ CLI Command Reference

🧠 1. Memory & Knowledge Ingestion

• devmind remember: Scans files, AST symbols, comments, and git history into local graph-vector memory.

  devmind remember                # Fast local mode (0 API calls, instant)
  devmind remember --incremental  # Only scan changed files in git diff
  devmind remember --deep         # Run LLM entity extraction
  

• devmind ask "<query>": Ask plain-English questions about codebase architecture.

  devmind ask "How does payment retry logic work on webhook timeout?"
  

• devmind chat: Start an interactive terminal REPL session to explore code.

  devmind chat
  

• devmind log "<decision>": Record an Architectural Decision Record (ADR) into persistent memory.

  devmind log "Chose SQLite/LanceDB locally to ensure zero infrastructure overhead."
  

🔬 2. Code Quality & Health Analysis

• devmind health: Scans the codebase for technical debt, complexity hotspots, code smells, dead imports, and test coverage gaps (100% offline).

  devmind health                        # Rich terminal dashboard
  devmind health --output report.md     # Export Markdown report
  devmind health --threshold 75         # CI quality gate (exit code 1 if score < 75)
  

🗺️ 3. Visual Graphs & Architecture Digests

• devmind graph: Launch an interactive browser graph of codebase symbols and relationships.

  devmind graph --port 8000
  

• devmind digest: Generate a structured Markdown architecture map of classes, functions, and files.

  devmind digest --output ARCHITECTURE_MINDMAP.md
  

• devmind dashboard: Open the local web management dashboard.

  devmind dashboard
  

⚙️ 4. Configuration & Maintenance

• devmind init / devmind config: Interactive setup wizard to switch providers or keys.
• devmind refresh: Re-scan changed files and refine relationship links in graph memory.
• devmind forget: Surgically remove a file or wipe local memory.

  devmind forget --file auth/middleware.py   # Delete specific file memory
  devmind forget --all                      # Wipe local memory databases
  

🤖 Agentic IDE Integration (Claude Code & Cursor via MCP)

Connect DevMind's persistent memory directly to autonomous agents like Claude Code, Cursor Agent, or Windsurf using the Model Context Protocol:

Claude Code Setup:

claude mcp add devmind "devmind mcp"

Cursor / Project .mcp.json Setup:

Create a .mcp.json file in your repository root:

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

Your AI assistant can now call query_codebase_memory and log_decision_record automatically instead of burning 50,000+ tokens blindly reading random files!

📄 License

DevMind is open-source software licensed under the MIT License.