MCP Memory-mesh

Tools · 15

• search_memory Hybrid search over all indexed content. Returns path, preview, score, file type, source, and optional extended_preview for wider context.
• list_sources List all configured sources with file counts and index status.
• get_document Read the full content of an indexed file (up to 1 MB by default).
• index_now Force immediate re-index of a file or directory, bypassing the watcher.
• ask_memory RAG: retrieves relevant passages and sends them to a local Ollama model for a grounded answer. Requires Ollama running locally.
• pin_memory Pin a chunk to the hot tier — never demoted, never score-decayed.
• forget_memory Suppress a chunk from future search results without deleting the source file.
• query_timeline Query the episodic event log: what was retrieved / indexed in the last N days?
• sync_source Pull and index documents from a configured external connector (Jira, Notion, GitHub…).
• get_entity Look up a named entity (person, project, concept) and its associated chunk IDs.
• related_documents Find documents semantically similar to the given file path.
• search_by_date Search indexed chunks by last-modified date range.
• forget_source Remove all indexed data for a named source from the index.
• summarize_source Generate a brief summary of the most recent content in a source (requires Ollama).
• graph_memory Return the entity co-occurrence knowledge graph as nodes and edges.