Context Portal MCP (ConPort)
@GreatScottyMac
Context Portal (ConPort): A memory bank MCP server building a project-specific knowledge graph to supercharge AI assistants. Enables powerful Retrieval Augmented Generation (RAG) for context-aware development in your IDE.
Overview
What is Context Portal MCP (ConPort)?
Context Portal MCP (ConPort) is a database-backed Model Context Protocol server that functions as a project's "memory bank" for AI assistants. It stores structured project context—such as decisions, tasks, and architectural patterns—in a SQLite database, enabling AI to retrieve relevant information via semantic search (vector embeddings) and power Retrieval Augmented Generation (RAG). It is designed for developer tools and IDEs that support MCP.
How to use Context Portal MCP (ConPort)?
Install via uvx (recommended) by adding a configuration entry to your MCP client settings (e.g., mcp_settings.json) with the command uvx and arguments including conport-mcp with --mode stdio and --workspace_id. Alternatively, clone the repository, create a virtual environment with uv venv, install dependencies, and run via uv run. Use one of the provided custom instruction strategy files (for Roo Code, Cline, Windsurf, or generic) to guide your LLM agent on how to use ConPort tools.
Key features of Context Portal MCP (ConPort)
- SQLite-backed structured context storage (one database per workspace)
- Multi-workspace support via
workspace_idparameter - Vector embeddings for semantic search and RAG capabilities
- Project knowledge graph with explicit relationships between context items
- Runs in STDIO mode for tight IDE integration
- Includes Alembic migrations for seamless schema updates
Use cases of Context Portal MCP (ConPort)
- AI assistants maintaining persistent project memory across sessions
- Developers storing and querying decisions, progress, and architectural designs
- Enabling AI to answer questions using up-to-date, project-specific information
- Building a knowledge base for sprint planning and retrospectives
