ClaudeMap

agent-memory

A long-term memory runtime for AI agents: a markdown store with local ranked retrieval, shared by Claude Code and Codex CLI, no API keys.

SDKs & Toolingmemoryretrievalmarkdownlocal-firstcli

agent-memory keeps markdown files in one store as the single source of truth, with a local SQLite index beside them as a cache you can delete at any time. Recall answers with file paths ranked by BM25 rather than pasted text, the agent opens each hit only as deep as the task needs, and a sleep-time pass consolidates and forgets by value on its own clock. Claude Code, Codex CLI, and anything that can run a shell command share the same store.

What it does

  • Markdown files are the source of truth; the SQLite FTS5 index is a deletable cache — nothing opaque to migrate off
  • Recall returns an L0 list (one-line abstract, path, anchor, score) and mem read opens it at outline / abstract / full depth
  • Three read tracks: deterministic MEMORY.md injection at session start, BM25 recall with an optional vector plugin fused by RRF, and plain ls/grep over the directory
  • Writes fire at conversation boundaries and a sleep-time pass consolidates and forgets by value — all local, no API key

When to use it

For anyone whose coding agent forgets decisions between sessions: keep project decisions, preferences, and post-mortems in one store that Claude Code and Codex CLI both read and write, so the next session starts warm instead of from zero.

Getting started

# Requires Python 3.12+ and uv; no PyPI release yet
git clone https://github.com/tigerless-labs/agent-memory.git
cd agent-memory
uv sync --all-packages

mem init
# store defaults to ~/agent-memory-store;
# export AGENT_MEMORY_STORE to relocate it — in every shell your agents run in

Things to know

Version 0.1.0 and not on PyPI yet, so you are installing from a checkout — expect churn. The store path must be exported consistently in every environment your agents run in, or each one quietly gets its own store.

Repository

tigerless-labs/agent-memory

Charted

Related on the map

  • Harness Bridge

    SDK / Tool

    A local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.

    • launcher
    • local-models
    • cli
    • menu-bar
  • Claude for Siri

    SDK / Tool

    Answers macOS Spotlight and Siri prompts with your signed-in Claude Code account — an experimental macOS 27 model-delegation provider.

    • macos
    • siri
    • spotlight
    • model-delegation
  • Extract your own local chat history from ten AI coding assistants into one normalized JSONL — for fine-tuning, analytics, or backing up years of conversations. Standard library only.

    • cli
    • data-export
    • jsonl
    • backup