DevContext: Autonomous Context Awareness Model-Context-Protocol (MCP) Server

What is DevContext?

DevContext is a Model Context Protocol (MCP) server that provides continuous, project-centric context awareness for developers. It uses non-vector retrieval methods—keyword analysis, relationship graphs, and structured metadata—to understand both your codebase and conversations, adapts to development patterns, and integrates with Cursor IDE. Data is stored in a TursoDB database dedicated to a single project.

How to use DevContext?

Install Node.js 18+, set up a TursoDB database, and configure the MCP server in your project’s .cursor/mcp.json with TURSO_DATABASE_URL and TURSO_AUTH_TOKEN environment variables. Run npx -y devcontext@latest. Optionally, implement the provided Cursor Rules for autonomous context management, which define a precise tool execution sequence: initialize_conversation_context, update_conversation_context, retrieve_relevant_context, record_milestone_context, and finalize_conversation_context.

Key features of DevContext

• Non-vector context retrieval (keyword, relationship graph, metadata)
• Project‑centric design with a dedicated database per project
• Minimal dependencies (MCP SDK, TursoDB client, lightweight AST parsing)
• Hierarchical understanding from repository structure to function/variable level
• Intelligent context prioritization using recency, importance, relationships, and focus

Use cases of DevContext

• Maintain full conversational context across a development session
• Retrieve relevant codebase context on demand (functions, files, architecture)
• Record significant milestones (features, bug fixes, architectural decisions)
• Enable autonomous context management and task workflows via Cursor Rules

FAQ from DevContext

What differentiates DevContext from vector‑based context systems?

DevContext uses keyword analysis, relationship graphs, and structured metadata instead of vector embeddings, avoiding external ML dependencies and providing language‑aware retrieval for multiple programming languages.

What are the runtime and database requirements?

Node.js 18 or higher, a TursoDB account (free tier works), and environment variables TURSO_DATABASE_URL and TURSO_AUTH_TOKEN.

Where does context data live?

All context, metadata, and optional logs are stored in a TursoDB database dedicated to the project. No data is stored locally outside the database.

What transport and authentication does DevContext use?

It communicates via MCP stdio (process‑based transport). Authentication to TursoDB is handled through the TURSO_AUTH_TOKEN environment variable.

Are there any known limits?

Each server instance is designed for a single project, eliminating cross‑project complexity. There is no cross‑project context sharing.