Memory Fort

Cross-tool persistent memory for AI agents — local, private, and free.

Memory Fort gives every AI coding session a shared long-term memory: observations flow in automatically from Claude Code, Codex, Antigravity, Hermes, Pi, and OpenClaw; a curated wiki of markdown pages grows over time; and retrieval (BM25 + semantic + graph) surfaces the right context at session start. No database. No external service. No API key to get started.

Your memory is a folder of plain text files — a git repo, an Obsidian vault, and a typed knowledge graph all at once.

Why Memory Fort?

Most agent memory tools require a cloud account, a running database, or a paid API to work at all. Memory Fort does not.

• Your data, your machine. Everything lives under ~/.memory/ as markdown files you can read, edit, grep, and version-control.
• No vendor lock-in. Open schema, plain text format, vault is just a git repo.
• No account required to start. Lexical search (BM25 + graph) works on day one with zero API keys.
• Obsidian-native. Open ~/.memory/ in Obsidian and get a knowledge graph, backlinks, and full-text search for free.
• Cross-tool hooks. Claude Code, Codex, Antigravity, Hermes, Pi, and OpenClaw all write to the same vault automatically — one memory across all your AI tools.

Quickstart

npx memory-fort init

Interactive wizard asks ≤4 questions (all pre-defaulted), detects your installed tools, and wires everything. Press Enter to accept all defaults.

Prerequisites: Node.js ≥ 20. Nothing else. No Docker, no database, no API key.

# Search immediately (no key needed)
memory-fort grep "your query"

# Browse and search in the UI
memory-fort dashboard

How it works

System architecture

flowchart TD
    subgraph Tools["AI Tools"]
        CC["Claude Code"]
        CX["Codex"]
        AG["Antigravity"]
        HE["Hermes"]
        PI["Pi / OpenClaw"]
        CD["Claude Desktop"]
        VS["VS Code"]
    end

    subgraph Capture["Capture (automatic)"]
        H["Hook scripts\nPreToolUse · PostToolUse\nUserPromptSubmit · Stop"]
        MCP_IN["MCP server\nlog_observation"]
    end

    subgraph Vault["~/.memory/ (vault)"]
        RAW["raw/YYYY-MM-DD/\nsession.md"]
        WIKI["wiki/\nprojects · decisions · lessons\ntools · threads · procedures"]
        IDX["index.md · schema.md\npreferences.md"]
        EMB["embeddings/\n*.jsonl sidecars"]
    end

    subgraph Curation["Curation"]
        COMPILE["memory compile\n(distills raw → wiki)"]
    end

    subgraph Retrieval["Retrieval — three routes"]
        SS["① Session-start hook\nautomatically at every session"]
        SEARCH["② MCP memory.search\nBM25 + vectors + graph + RRF"]
        GREP["③ memory grep\nripgrep over raw + wiki"]
    end

    CC & CX & HE & PI -->|hooks fire| H
    AG & CD & VS & PI -->|MCP calls| MCP_IN
    H --> RAW
    MCP_IN --> RAW
    RAW --> COMPILE
    COMPILE --> WIKI
    WIKI --> IDX
    IDX --> SS
    WIKI & RAW & EMB --> SEARCH
    RAW & WIKI --> GREP
    SS -->|context injected at session start| Tools
    SEARCH -->|ranked results via MCP| Tools
    GREP -->|matching lines via CLI| Tools

How memories reach your AI tools

flowchart LR
    subgraph V["~/.memory/"]
        S["schema.md"]
        I["index.md"]
        L["log.md"]
        P["wiki/preferences.md"]
        W["wiki/projects/<cwd>.md"]
        WK["wiki/ pages"]
        EM["embeddings/"]
        R["raw/"]
    end

    subgraph Routes["Retrieval routes"]
        R1["(1) Session-start\nautomatic push\nevery session"]
        R2["(2) MCP search\non-demand pull\nBM25 + vectors + graph"]
        R3["(3) grep\nCLI / manual"]
    end

    subgraph AI["AI Tool"]
        CTX["Session context\n(prompt prefix)"]
        TOOL["Agent tool call\n(MCP response)"]
        CLI2["CLI output"]
    end

    S & I & L & P & W -->|reads on session start| R1
    WK & EM & R -->|hybrid retrieval + RRF| R2
    R & WK -->|ripgrep| R3

    R1 -->|stdout injected by tool| CTX
    R2 -->|ranked snippets| TOOL
    R3 -->|matching lines| CLI2

Route (1) fires automatically — you always get your top context injected. Routes (2) and (3) are on-demand (agent or human asks).

Supported tools

memory-fort install claude-code     # Claude Code (full hooks + plugin)
memory-fort install codex           # Codex desktop + CLI (hooks + MCP)
memory-fort install antigravity     # Google Antigravity / Gemini (MCP + live-capture plugin)
memory-fort install hermes          # Hermes agent (YAML hooks + MCP in ~/.hermes/config.yaml)
memory-fort install pi              # Pi coding agent (YAML hooks in ~/.pi/config.yaml)
memory-fort install openclaw        # OpenClaw (MCP server in ~/.openclaw/openclaw.json)
memory-fort install claude-desktop  # Claude Desktop (MCP only)
memory-fort install vscode          # VS Code (MCP only)

All installs are non-destructive and idempotent — sentinel-block writes, re-running is safe.

# Undo any integration cleanly
memory-fort uninstall claude-code
memory-fort disconnect --all

Retrieval modes

| Mode | Needs | When to use | |---|---|---| | Lexical (default) | Nothing | Day 1, offline, private projects | | Voyage embeddings | VOYAGE_API_KEY | Best semantic recall | | OpenAI embeddings | OPENAI_API_KEY | Alternative to Voyage | | Ollama (local) | Ollama running locally | Full local, no cloud at all |

Switch any time: edit ~/.memory/config.yaml or re-run memory-fort init.

Memory Fort vs. other agent memory tools

All claims below are sourced from 2026 benchmarks and vendor documentation.

| | Memory Fort | mem0 | Zep / Graphiti | Letta | Cognee | LangMem | OMEGA | |---|---|---|---|---|---|---|---| | Storage | Markdown files | Cloud DB / OSS | Cloud only¹ | PostgreSQL | SQLite + LanceDB | You choose | SQLite | | Requires API key | ❌ No (lexical default) | ✅ Yes | ✅ Yes | ✅ Yes (LLM) | ✅ Yes (LLM) | ✅ Yes (LLM) | ❌ No | | Self-hosted | ✅ Always | ✅ OSS option | ❌ Dropped¹ | ✅ Free | ✅ Local | ✅ OSS | ✅ | | Offline / air-gapped | ✅ | ❌ | ❌ | ❌ | ✅ (local LLM) | ❌ | ✅ | | Human-readable | ✅ Markdown + YAML | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | | Obsidian-compatible | ✅ Native | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | | Git-backed | ✅ Built-in | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | | Multi-tool hooks | ✅ 6 tools | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | | Knowledge graph | ✅ Typed edges (free) | ✅ Pro only ($249/mo) | ✅ Graphiti | ✅ All tiers | ✅ All tiers | ❌ | ❌ | | LongMemEval | — | 49.0%² | 63.8%² | — | — | 59.8s p95 latency³ | 95.4%⁴ | | Free tier | Unlimited (local) | 10K memories, 1K calls/mo | ❌ | ✅ self-hosted | ✅ self-hosted | ✅ OSS | Unlimited | | TypeScript SDK | ✅ Native CLI + MCP | ✅ | ✅ | partial | ❌ Python only | partial | ❌ |

¹ Zep dropped its self-hosted Community Edition in 2025; Zep Cloud is now the only supported path.
² Agent Memory at Scale 2026 — AgentMarketCap
³ Best AI Agent Memory Frameworks 2026 — Atlan
⁴ OMEGA comparison page

When to choose something else

• mem0 — managed cloud, polished dashboard, widest language support, largest community.
• Zep / Graphiti — best temporal fact tracking; reason about when facts changed.
• Letta — full stateful agent runtime, not just memory.
• Cognee — rich multimodal pipeline (images, audio, documents); Python-only.
• LangMem — natural fit on LangChain/LangGraph; note the high p95 latency for interactive use.
• OMEGA — fully local + AES-256 encryption at rest; no multi-tool hook support.

Wiki schema

Memory Fort organizes curated knowledge by entity type:

| Type | Directory | Purpose | |---|---|---| | projects | wiki/projects/ | Codebases and work efforts | | decisions | wiki/decisions/ | Architecture and tooling choices, with alternatives | | lessons | wiki/lessons/ | Reusable facts learned from incidents | | references | wiki/references/ | Papers, posts, talks | | tools | wiki/tools/ | Libraries and services | | threads | wiki/threads/ | Narrative arcs across a stretch of work | | procedures | wiki/procedures/ | Reusable step-by-step workflows |

Pages link via typed graph edges (uses, depends_on, supersedes, contradicts, caused_by, fixed_by, derived_from). Plain YAML frontmatter — no database required.

Dashboard

memory-fort dashboard
# → http://127.0.0.1:4410/memory/

Built-in React dashboard: browse the wiki, search (BM25 + semantic + graph), review proposed pages, inspect graph health metrics.

Roadmap

• OpenCode integration (memory-fort install opencode) — plugin drop into ~/.config/opencode/plugins/
• Optional SQLite-FTS index — rebuildable cache for sub-10ms lexical search at large vault sizes
• Community integrations — pull requests welcome; hook pattern documented in docs/architecture.md

License

Memory Fort is source-available under the PolyForm Noncommercial License 1.0.0 — free for personal use, hobby projects, and non-commercial research.

Commercial use requires a paid license. See COMMERCIAL.md.

For contributors / private dev repo

After cloning, install the pre-push gate:

npm run install:dev-hooks

This gates every git push origin through scan:leaks so personal tokens can never accidentally reach the public repo.

Built by GalaxyRuler