neverinfamous-memory-journal-mcp — Claude Code MCP server
Keeps a running record of project work so a fresh AI session knows what came before. Decisions, fixes and past mistakes stay searchable.
claude mcp add --transport http neverinfamous-memory-journal-mcp https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcpWhat neverinfamous-memory-journal-mcp does
The server gives your agent a long-term record of a codebase. Entries are stored in a database searched two ways at once: keyword matching through FTS5 plus vector similarity, merged with reciprocal rank fusion. Ask why a service ended up on SQLite rather than Postgres a month back, and the agent looks it up instead of guessing.
GitHub work sits alongside the journal: issues, pull requests, Actions, Kanban, milestones and reviews. There are workflow prompts too, and a graph linking decisions, specs and PRs that can be drawn with Mermaid. Code Mode runs multi-step operations through an mj.* API; the author claims token savings up to 90%, which is worth measuring on your own repos.
Written in TypeScript, shipped on npm and as an Alpine Docker image. The setup shown below points at a remote HTTP endpoint, so nothing needs installing locally for that route. Run it yourself when the journal should stay with you: the database is shared, and entries are visible to teammates. GitHub tools need you to be authenticated with GitHub first; without that they will not work.
What it helps with
- recover why one library was picked over another weeks ago
- hand working context to the next AI session without copying it by hand
- close an issue and log the reason behind an architectural fix
- trace how a spec connects to its implementation and pull request
- flag a blocker for a teammate so their agent picks it up
Example prompts for the agent
- Why did we go with SQLite instead of Postgres for this service last month?
- Close issue #42 and write down how the parsing bug was fixed.
- Raise a blocker flag for @sarah — I'm stuck on a database error.
How to install neverinfamous-memory-journal-mcp
Platform CLI: remote server
claude mcp add --transport http neverinfamous-memory-journal-mcp https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcpPlatform CLI: remote server
codex mcp add neverinfamous-memory-journal-mcp --url https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcpConfig file
{
"mcpServers": {
"neverinfamous-memory-journal-mcp": {
"url": "https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcp"
}
}
}Config file
{
"servers": {
"neverinfamous-memory-journal-mcp": {
"type": "http",
"url": "https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcp"
}
}
}Platform CLI: remote server
gemini mcp add -t http neverinfamous-memory-journal-mcp https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about neverinfamous-memory-journal-mcp
How do I install neverinfamous-memory-journal-mcp in Claude Code?
Run claude mcp add --transport http neverinfamous-memory-journal-mcp https://server.smithery.ai/@neverinfamous/memory-journal-mcp/mcp. That wires up the hosted HTTP endpoint; no local install is involved.
What does neverinfamous-memory-journal-mcp do?
It keeps a searchable journal of past project work that the agent can query, and adds GitHub tools for issues, pull requests, Actions and milestones.
How is it different from keeping notes in the chat?
The journal is queried by meaning, not just exact words, because keyword and vector search are combined. Briefings also include live CI health and local Git status.
