Repomemory
Persist what your coding agent learned across sessions with hybrid search, auto capture, and context routing for the repo.
Overview
RepoMemory MCP is a MCP server for the Build phase that gives coding agents persistent hybrid-search memory with auto-session capture and context routing.
What is this MCP server?
- npm package repomemory version 1.3.0 with stdio transport
- Hybrid search across stored agent and repo session memory
- Auto-session capture so decisions survive thread boundaries
- Context routing to surface the right memories per task
- GitHub repository DanielGuru/repomemory linked in server manifest
- Package version 1.3.0
- Three described mechanisms: hybrid search, auto-session capture, context routing
- stdio npm identifier repomemory
What problem does it solve?
Stateless agent chats forget yesterday's architecture calls, repro steps, and half-done refactors.
Who is it for?
Indie developers running long-horizon Claude Code or Cursor work on one repo who need durable session memory.
Skip if: One-shot snippet generation with no expectation of cross-session continuity.
What do I get? / Deliverables
After registration, sessions accumulate searchable memory and routed context so the agent continues work without full re-briefs.
- Searchable persistent memory across agent sessions
- Automatic capture hooks for session outcomes
- Routed context bundles for task-specific prompts
Recommended MCP Servers
Journey fit
Memory MCP is first installed when you start relying on agents to implement features and need continuity, even though it helps later phases too. agent-tooling is the shelf because RepoMemory extends the assistant runtime with memory primitives, not a one-off deploy or marketing integration.
How it compares
Persistent agent memory MCP, not a static vector doc upload or live semantic code graph alone.
Common Questions / FAQ
Who is RepoMemory MCP for?
Solo builders and agent-first teams who work on the same codebase across many sessions and need captured context to stay searchable.
When should I use RepoMemory MCP?
Use it as soon as agent-assisted implementation becomes your default workflow and you feel repetition cost from missing session history.
How do I add RepoMemory MCP to my agent?
Install the repomemory npm package, configure it as a stdio MCP server in your client, and enable session capture and hybrid search in your project workflow.