Context Mem
Give your coding agent durable memory across sessions with semantic retrieval and a large MCP tool surface to cut repeated context stuffing.
Overview
io.github.JubaKitiashvili/context-mem is a MCP server for the Build phase that provides persistent agent memory with semantic retrieval across forty-four MCP tools.
What is this MCP server?
- 44 MCP tools for memory storage, retrieval, and management per server description
- Vendor-claimed 98%+ retrieval recall and 99% token savings in catalog listing
- Stdio serve via npx -y context-mem serve at version 3.2.1
- Persistent cross-session memory aimed at long-horizon solo-builder agent work
- 44 tools stated in server description
- Catalog claims 98%+ retrieval recall and 99% token savings
- npm package context-mem version 3.2.1 with stdio transport
What problem does it solve?
Long agent threads lose project context after compaction or new sessions, forcing you to re-upload specs and repeat explanations.
Who is it for?
Solo builders on week-long agent-led builds who need durable memory and many granular memory MCP operations.
Skip if: One-shot tasks with no return visits, or teams that already centralize memory in a managed SaaS with no local MCP need.
What do I get? / Deliverables
After you run context-mem serve, the agent can store and recall structured memories so later phases reuse facts with less prompt bloat.
- Cross-session memory stores the agent can query via MCP
- Structured recall workflows using the advertised 44-tool surface
- Lower repeated context in prompts when retrieval replaces full history paste
Recommended MCP Servers
Journey fit
Agent-tooling on the build shelf is where persistent memory is first adopted, even though recall helps every later phase. context-mem is infrastructure for agent sessions—stores, search, and memory ops—not app frontend or SEO.
How it compares
Dedicated agent memory MCP with dozens of tools, not a simple notes skill or vector DB you wire yourself.
Common Questions / FAQ
Who is io.github.JubaKitiashvili/context-mem for?
Indie builders and agent power users who run Claude Code, Cursor, or Codex across many sessions and need persistent, searchable project memory.
When should I use io.github.JubaKitiashvili/context-mem?
Use it from early build through operate whenever context window limits or session resets would otherwise erase decisions, APIs, or customer details.
How do I add io.github.JubaKitiashvili/context-mem to my agent?
Add a stdio MCP server with runtime hint npx -y context-mem serve, ensure Node.js is installed, then connect from your MCP client per context-mem repository docs.