Mcp External Memory
Persist decisions, user prefs, and project facts outside the chat window so your agent can search semantic memory across sessions.
Overview
io.github.daedalus/mcp-external-memory is an MCP server for the Build phase that provides persistent semantic memory tools for coding agents.
What is this MCP server?
- Persistent, searchable semantic memory for LLM workflows
- MCP server mcp-external-memory v0.1.2 on PyPI with stdio
- Survives session resets—store notes, architecture, and customer context
- Semantic search instead of dumping entire repos into context
- Fits multi-week indie builds where the agent must remember prior choices
- Server version 0.1.2
- Transport: stdio
- PyPI identifier: mcp-external-memory
What problem does it solve?
Long agent sessions forget decisions and user context, forcing you to re-explain architecture and preferences every day.
Who is it for?
Indie builders running multi-session agent workflows who need durable recall for specs, stack choices, and customer support notes.
Skip if: One-shot codegen tasks or teams that already standardize on a single hosted memory SaaS with its own native integration.
What do I get? / Deliverables
The agent stores and retrieves semantic memories via MCP so project knowledge compounds across build and operate cycles.
- MCP tools to write and query persistent semantic memories
- Cross-session continuity for specs, prefs, and operational notes
Recommended MCP Servers
Journey fit
External memory is shelved under Build agent-tooling because you register it while wiring the agent stack, even though you reuse it when operating and growing the product. Agent-tooling captures MCP servers that extend recall and context—core infrastructure for long-horizon solo shipping.
How it compares
Agent memory MCP layer, not a vector database admin UI or generic note-taking skill.
Common Questions / FAQ
Who is io.github.daedalus/mcp-external-memory for?
Solo builders and small teams using MCP agents who need searchable memory that outlives a single chat session.
When should I use io.github.daedalus/mcp-external-memory?
Use it when you start a multi-day product build or support loop and want the agent to recall prior decisions without huge context pastes.
How do I add io.github.daedalus/mcp-external-memory to my agent?
Install mcp-external-memory from PyPI, configure stdio MCP in your client, set any storage or embedding env vars from the repo README, then restart.