heimdall-mcp-server

Heimdall MCP Server solves the problem of AI coding assistants starting each session with no memory of your codebase or previous conversations. It maintains a persistent, project-specific knowledge base by indexing your documentation, git history, and development insights, then makes this context available through the Model Context Protocol. The system uses Qdrant vector search and SQLite metadata storage to efficiently retrieve relevant memories when your LLM queries the system. Developers and operations teams benefit from an AI assistant that understands architectural patterns, past solutions, and project-specific context across multiple sessions.

[{"step":"Define a clear label for your context (e.g., 'project_config_v1'). Use concise, descriptive names like 'api_endpoints_2024' or 'debugging_steps_issue_123'.","tip":"Labels should be unique and avoid spaces or special characters. Prefix with a domain like 'frontend_' or 'backend_' for better organization."},{"step":"Store context using the heimdall-mcp-server command: 'Store the following context with label [LABEL]: [CONTENT]'. Include all relevant details like versions, configurations, or troubleshooting steps.","tip":"For large contexts (e.g., entire project configs), break them into smaller chunks and store them under related labels (e.g., 'frontend_deps', 'backend_deps')."},{"step":"Retrieve context in future sessions using: 'Retrieve the context labeled [LABEL]'. The server will return the stored data in a structured format.","tip":"Use the 'list' command to see all available labels if you forget the exact name. Example: 'List all stored context labels'."},{"step":"Update context by reusing the same label with new content. The server will overwrite the previous entry, ensuring you always have the latest information.","tip":"Add a 'last_updated' timestamp and brief notes to track changes over time. Example: 'Added axios for API calls; removed deprecated request library'."},{"step":"Integrate with your workflow by automating context storage. For example, use a pre-commit hook to store the latest project state before pushing changes.","tip":"Combine with CI/CD tools like GitHub Actions to ensure context is stored alongside code changes. Example: 'Store context with label project_state_after_commit_[SHA]'."}]