Memoir
Persist structured session journals so your coding agent remembers decisions, blockers, and outcomes across days and repos.
Overview
Memoir is an MCP server for the Build phase that maintains structured session journals so AI agents retain persistent memory across sessions.
What is this MCP server?
- Structured session journals designed for AI agents with fields for outcomes and resolutions
- Rolling limits configurable via MEMOIR_MAX_ENTRIES, MEMOIR_MAX_SESSIONS, MEMOIR_MAX_CONTENT, MEMOIR_MAX_OUTCOME
- stdio npm package memoir-mcp v1.0.3 for local, project-scoped memory
- Targets persistent memory across sessions rather than one-shot summarization
- Server version 1.0.3 on npm identifier memoir-mcp
- Four documented env vars: MEMOIR_MAX_CONTENT, MEMOIR_MAX_OUTCOME, MEMOIR_MAX_ENTRIES, MEMOIR_MAX_SESSIONS
What problem does it solve?
Long-running agent work resets every new chat, so you repeat explanations and lose the narrative behind technical decisions.
Who is it for?
Solo builders who rely on Claude Code or Cursor for multi-session features, refactors, or incident response on one codebase.
Skip if: Teams that need enterprise knowledge search across wikis—Memoir is local structured journals, not a company-wide RAG corpus.
What do I get? / Deliverables
After you add Memoir, agents can read and append structured session history so follow-up work continues with explicit prior context.
- Structured journal entries per agent session
- Rolling project-scoped session history the agent can query on restart
Recommended MCP Servers
Journey fit
Build is where long agent sessions accumulate context; Memoir is agent-tooling that keeps that context durable instead of lost when the chat ends. Agent-tooling is the right shelf for memory and journaling primitives that every phase can reuse once installed, even though operate benefits too.
How it compares
Persistent agent session memory MCP, not a vector database or generic note-taking plugin.
Common Questions / FAQ
Who is Memoir for?
Indie developers using MCP agents who want durable, structured session logs tied to a project instead of fragile chat scrollback.
When should I use Memoir?
Use it when sessions span days, you hand off context between chats, or you want the agent to record decisions and resolutions as you build.
How do I add Memoir to my agent?
Register memoir-mcp (v1.0.3) over stdio in your MCP client and optionally tune MEMOIR_MAX_CONTENT, MEMOIR_MAX_OUTCOME, MEMOIR_MAX_ENTRIES, and MEMOIR_MAX_SESSIONS to fit your machine.