Anamnesis plugin

Local memory layer: imports Claude Code, Codex and mem0 memory into SQLite and serves it back to agents over CLI and MCP.

  • 188 ★ on the repository
  • Last commit
  • Apache-2.0
Plugin

Extended access

Why this level

  • Reads local sessions and memory of several agents, which may hold secrets and personal data.
  • Admin import over MCP opens write access to the store.
All reasons and checks

trapezohe/anamnesis

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add plugins/anamnesis

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Reads local sessions and memory of several agents, which may hold secrets and personal data.
  • Admin import over MCP opens write access to the store.

This is third-party code. Review the repository files before installing.

What it does

Anamnesis is an interoperability layer between agent memory stores. It reads what agents have already accumulated: Claude Code sessions, Codex memory, mem0 and Letta databases, OpenClaw workspaces and the resources of any MCP server. Records are normalized into a shared schema with source, scope and timestamps, indexed in local SQLite through FTS with optional embeddings, and served by search, export and the MCP tools `search`, `read`, `list`, `doctor` and `watch_status`. The original agent still creates the memory; Anamnesis only gathers and serves it. The `anamnesis watch` daemon keeps the store fresh and can start at login. The Claude Code plugin adds an MCP config, the `anamnesis-memory` skill and the `/anamnesis-status` and `/anamnesis-search` commands; Codex connects through `codex mcp add`. Import over MCP is hidden and enabled only by an admin flag.

Who it is for. People working in several agents at once who want Claude Code to see what Codex or mem0 accumulated without adopting yet another memory system.

Good fit when

  • Months of context from Codex need to be pulled into Claude Code.
  • mem0 and Letta records are searched with one query alongside agent sessions.
  • The memory of an external MCP server is registered as another source.
  • The store refreshes in the background whenever source files change.

Not a fit when

  • There is only one agent and its built-in memory is enough.
  • The source has no adapter: parsing precision varies.
  • Cloud sync without a local SQLite store is needed.

Example request

Search the shared memory for every note about the database migration and show which agent each came from.

Limitations

Building needs Rust 1.85 or newer, and the binaries are installed with `cargo install`. “Supported” means an adapter and an import command exist; parsing precision varies, medium for Claude Code and Codex and high for mem0 when the schema matches. Embeddings are local and deterministic by default, with an external provider as a separate option. The repository belongs to the Ghast team and the project is developed as part of its workspace, so adapter upkeep depends on them.

How to disable. Remove the plugin through the client's plugin manager or `codex mcp remove anamnesis`. Stop `anamnesis watch`, remove the launchd or systemd autostart and delete the local database.

FAQ

Which sources does Anamnesis read?

Claude Code, Codex, Hermes, OpenClaw, mem0, Letta, TDAI, OpenViking, MemPalace, Memori, MemOS, Memary and any MCP server through `generic-mcp`. `anamnesis discover` probes known local paths.

How do I connect it to Codex?

Run `codex mcp add anamnesis -- anamnesis serve`. Do not enable direct MCP and the marketplace plugin together: registrations would duplicate.

Why is import unavailable over MCP?

Import is an admin operation and stays hidden unless the server starts with admin tools enabled. Read and search are the default.

Where is the data stored?

In local SQLite: metadata, chunks, audit events and lineage. The agents' original stores stay where they are and are not modified.

What does search offer?

Keyword, vector and hybrid search with filters by source, instance, kind, scope and time. Embeddings are local and deterministic by default.

Discussion

No questions or reviews yet.

Plugins

Shared memory for coding agents: messages in SQLite, old context compacted into a summary tree and MCP tools to search it.

PluginExtended access60Repository stars
Official

Local persistent memory for coding agents: hooks record every session, and Claude Code, Codex or Cursor recall past decisions through 54 MCP tools

MCP serverExtended access30.3KRepository stars

A long-term memory plugin that keeps decisions, preferences and task state across sessions and injects only what a query needs

PluginBroad access401Repository stars

Shared local memory for AI agents: Claude Code, Codex, Cursor and others find the agreements of past sessions in it and record new ones

CLIBroad access2.2KRepository stars

One local memory shared by Claude Code, Codex and OpenClaw: tell one agent, another knows tomorrow, no cloud or keys.

PluginExtended access111Repository stars

Comparison with similar tools

Cards that share the main search query, and how they differ.

ToolOfficialAuthenticationRepository starsMain difference
Anamnesis (this card)no188
agentmemoryyesnot required30,340records session progress through hooks and returns it to agents; Anamnesis imports existing memory from other systems
MCP Memory Servicenonot required2,117stores long-term memory itself with a knowledge graph; Anamnesis acts as a bridge between stores

README in short

Adapters for a dozen memory stores and one local index with MCP, instead of a new system where everything must be moved by hand.

README badge

Report a problem