Server facts
context-mem MCP Server
Project
- Author
- JubaKitiashvili
- Repository
- github.com/JubaKitiashvili/context-mem
- License
- MIT
- Category
- Memory & Knowledge
Tech
- Runtime / language
- node
- Transports
- stdio
- Remote available
- No
- Authentication
- Not specified
Activity
- GitHub stars
- 19
- Last commit
- Data as of
Overview
Coding decisions, debugging evidence, and session history become searchable project memory in context-mem. Observations are summarized according to their content type, stored in SQLite, and reflected in a linked Markdown vault that can be read independently of the MCP interface. Agents can save new observations, index material for later lookup, search across stored knowledge and content, or inspect a chronological history. Knowledge entries support contradiction detection and resolution, while graph tools connect entities and expose their relationships. Temporal queries retrieve facts valid at a chosen time, and decision explanations reconstruct the supporting evidence.
For ongoing coding work, session primers surface project context, handoffs save state with a continuation prompt, and restoration tools reload earlier snapshots. Budget controls and adaptive compression manage accumulated output while pinned entries remain verbatim. Session records can also become pull request narratives, standup updates, architecture decision records, or onboarding material. Multi-agent workflows include named registration, file claims with conflict reporting, and broadcasts. Retrieval combines lexical and vector methods, with optional LLM reranking; the underlying memory store and Markdown vault are local.
context-mem setup notes
- context-mem records stdio only; confirm a network transport upstream before planning remote access.