It allows AI coding tools (Claude Code, Cursor, OpenClaw/Claw Bot, etc.) to share the same project context in real-time, bridging the gap between coding and engineering.
Nous is particularly optimized for Autonomous Mobile Agents (like OpenClaw) that operate via messaging channels where project continuity is critical.
- Persistent Context: Share architectural decisions and code conventions across different AI tools.
- Auto-Indexing: Automatically builds a semantic knowledge graph of your repository.
- Standardized MCP Interface: Compatible with any tool that supports the Model Context Protocol.
- Decision Tracking: Record the "why" behind your code, not just the "what".
- Local-First: Privacy-focused storage using SQLite.
# Initialize Nous in your project
npx nous init
# Index your repository
npx nous indexnous init: Initializes the.nousdirectory and database.nous index: Scans the repository to build module and file context.nous status: Shows the current status of the knowledge base.nous decisions: Lists all recorded architectural decisions.nous serve: Starts the MCP server (stdio transport).
Nous exposes several tools to your AI agent:
get_project_context: Retrieve full context.record_decision: Log architectural choices.list_decisions: Filter and view decision history.get_module_overview: Get metadata for specific modules.record_convention: Store coding patterns.search_context: Fast keyword search.
- Phase 1: Core MCP Server & SQLite Schema.
- Phase 2 (In Progress): File-level indexing, Git hooks, and Semantic Search (Embeddings).
- Phase 3: Static exports (
CLAUDE.md,.cursorrules) and Git commit anchoring.
Open-core. MIT Licensed for the local server.