memd — Agent-Driven Project Memory Curator

memd gives your AI coding tools a memory that survives between sessions. It runs quietly in the background, reads the session logs your CLIs leave behind (claude-code, antigravity-cli, custom agents), and boils them down into a few small, git-versioned markdown files under ./.memory/ — the stuff worth remembering, minus the noise.

---

Why memd?

Every AI coding session ends the same way: the context window fills up, the session closes, and everything the model figured out — the port that service actually runs on, why you picked SQLite, the bug it hit last Tuesday — is gone. Next session it asks the same questions and steps on the same rakes.

memd fixes that by keeping a small set of memory files up to date automatically:

• Remembers what matters: current system state, active decisions, and open todos, in plain markdown you can read yourself.
• Skips what doesn't: conversational back-and-forth and tool spam get dropped; only durable facts survive.
• Works with a crowd: a lock-safe inbox lets any agent, script, or human hand a note to the curator without stepping on anyone else.
• Keeps receipts: memory changes are committed to git, so you can see how your project's context evolved over time.

---

How It Works

Sessions end (or a timer fires), memd collects whatever is new — transcripts, databases, inbox notes — scrubs it for secrets, and hands it to a small "curator" model that rewrites the memory files. Python code, not the model, gets the final say on what's allowed to change.

graph TD
    subgraph Clients / CLI Sessions
        CC[claude-code] -->|hooks / SessionEnd| MH[memd hook]
        AG[antigravity-cli] -->|SQLite DB/protobuf| DB[(conversations/*.db)]
        SA[Swarm/Agents] -->|Inbox Protocol v1.0| IN[(.memory/inbox/*)]
    end

    subgraph memd Core
        MH -->|Trigger| MS[memd sync]
        T[systemd sweep timer] -->|memd sweep| MW[Sweep Worker ThreadPool]
        MW -->|Sync Project| MS
        DB -.->|Parsed by| MS
        IN -.->|Ingested by| MS
    end

    subgraph Distillation & Curators
        MS -->|Assemble Prompt & Digest| PR[Redaction & De-duplication]
        PR -->|Prompt / Stdin| CB{Curator Backend}
        CB -->|Headless claude -p / Custom LLM| MS
        MS -->|Validate Outputs & Budgeting| VAL[Hard Invariants Enforcer]
        VAL -->|Write & Commit| MC[(.memory/ files)]
        VAL -->|Prune Overflow| ARC[(archive/YYYY-MM.md)]
    end

Loading

---

The Four Memory Files

Everything lands in four markdown files under ./.memory/:

File	What's in it	House rules
state.md	What's true right now: ports, directory layout, running services, active workarounds.	Present tense only. When a fact goes stale it gets replaced — no history kept here.
decisions.md	What was decided, why, and what that rules out.	Each entry has to actually constrain future work, not just describe the past.
todo.md	What's still open: tasks, roadmap items, things to verify.	Finished or abandoned items move to the archive.
mistakes.md	What went wrong before, and how to not do it again.	Append-only. Each entry: symptom, root cause, and the rule that prevents a repeat.

Anything that overflows the size budgets gets moved to ./.memory/archive/YYYY-MM.md instead of deleted.

---

Teaching Your Agents to Use the Memory

Keeping the memory curated is only half the job. The models working in your codebase won't read .memory/ unless something tells them it exists and how to behave around it. The rules boil down to three:

1. Read first. Check state.md, decisions.md, mistakes.md, and todo.md (or run memd brief for a short digest) before doing real work.
2. Don't edit the memory files. The curator owns them and rewrites them on every pass — hand-edits just get overwritten.
3. Send new facts through the inbox. memd note -m "...", or drop a file in .memory/inbox/. One fact per note.

You don't have to write those instructions yourself. contrib/agents/ ships ready-made instruction files for the major tools — drop the right one into your project and you're done:

Tool	File	Goes in
Codex, Cursor, Gemini CLI, Zed, and anything else on the agents.md standard	AGENTS.md	project root
Claude Code (knows about memd's hooks)	CLAUDE.md	project root
Gemini / antigravity-cli	GEMINI.md	project root
GitHub Copilot	copilot-instructions.md	.github/
Cursor (rules format)	memd.mdc	.cursor/rules/

Already have a CLAUDE.md or AGENTS.md? Append the memd section instead of overwriting — the files are written to compose. Details in contrib/agents/README.md.

---

Guardrails Enforced by Code (Not the LLM)

The curator model is useful but not trusted. The rules that keep memory intact live in plain Python, so a confused (or prompt-injected) model can't wreck the store:

1. Frontmatter required: Every memory file must carry valid YAML frontmatter (type, project, last_updated, status), or the write is rejected.
2. Mistakes are forever: mistakes.md is append-only. The curator can add entries but never edit or delete old ones.
3. Shrink guard: A distill that would delete more than 60% of a file gets rejected — unless the removed sections were explicitly moved to the archive.
4. Size budgets: Each file has a character cap (set in config.json). Overflow gets moved to ./.memory/archive/YYYY-MM.md, oldest sections first.
5. No races: Per-project file locking (flock) keeps concurrent sweeps, hooks, and agents from corrupting each other's writes.
6. Nothing skipped: memd tracks how far it has read into each transcript and database, and only advances that marker after a successful distill and write — a failed run just gets retried from the same spot.

---

Where the Raw Material Comes From

memd pulls from several sources before each distill:

1. claude-code Hooks

memd plugs into the claude-code CLI through lifecycle hooks in ~/.claude/settings.json (wired up for you by memd install-hooks):

• Session start: the memory brief gets injected into the assistant's starting context, so it begins the session already knowing the project.
• Session end / pre-compact: a background memd sync digests the transcript of the session that just finished.

2. antigravity-cli Database Parsing

For antigravity-cli, memd reads the SQLite conversation databases in ~/.gemini/antigravity-cli/conversations/*.db directly:

• Pulls out the useful steps — user inputs, assistant responses, tool actions, errors.
• Figures out which project a conversation belongs to by counting workspace-path mentions in the payloads (index kept in ~/.local/state/memd/ag_index.json).

3. The Inbox Protocol (v1.0)

The inbox is how everything else — another agent, an MCP tool, a CI script, you with a text editor — hands a fact to the curator without touching the memory files directly. Drop a markdown note in the inbox; the next sweep folds it into memory and deletes the note.

• Project Inbox: <project-root>/.memory/inbox/ (feeds local project files)
• Global Inbox: <global_root>/.memory/inbox/ (default ~/.memory/inbox/, feeds system-wide/cross-project files)

Rules for Writers

The inbox has many writers and one reader that deletes what it consumes, so a sloppy writer can lose data for everyone. Every writer must publish atomically:

1. Stage Outside the Inbox: Write the markdown note to a temporary file in the parent .memory/ directory (e.g. .remember-*.tmp).
2. fsync the File: Durably sync the data to disk (flush + fsync).
3. Atomic Publish: Publish to the inbox/ directory via rename(2) or link(2) (os.replace or os.link).
4. fsync the Inbox Directory: Ensure the parent directory structure is updated.
5. Collision-Proof Names: Name files using microsecond-resolution timestamp + PID (e.g., 20260711T123000123456-9876.md).
6. Write-Once: Never modify or reuse a published note.

For complete writer/reader guidelines, see INBOX-PROTOCOL.md.

4. Extra Custom Sources

Additional text transcript sources (e.g. custom JSONL logs) can be registered per-project under the projects.<path>.extra_sources array in config.json.

---

Configuration (config.json)

Configuration is stored in $XDG_CONFIG_HOME/memd/config.json (defaults to ~/.config/memd/config.json).

Key Reference & Defaults

Option	Type	Default	Description
claude_bin	string	"claude"	Name or path of the claude-code binary.
curator_cmd	array	[]	Override list of arguments to invoke a custom curator backend (e.g., ["nix", "run", ".#", "--", "status"]). Substitutes {model}. If empty, falls back to the headless claude_bin command.
antigravity_dir	string	"~/.gemini/antigravity-cli"	Directory containing native SQLite database conversations.
model_small	string	"haiku"	Model used for small/routine transcript distills.
model_large	string	"sonnet"	Model used for large/complex session distills.
escalate_chars	integer	15000	Character count above which digests are run with model_large.
digest_cap_chars	integer	60000	Maximum length of transcript digest fed to the LLM.
quiet_seconds	integer	600	Skip digesting transcripts modified within this cooldown period.
sweep_jobs	integer	4	Number of worker threads for parallel sweeps.
auto_scaffold	boolean	true	Auto-create .memory/ structure inside detected git repositories.
git_commit	boolean	true	Automatically run git commit on memory files after successful distill.
budgets	object	See below	Dictionary specifying character size caps for active memory files before archiving.
REDACT_EXTRA_PATTERNS	array	[]	List of raw regex patterns for custom credential redaction.
exclude	array	[]	List of absolute project paths to ignore during sweeps.
projects	object	{}	Mapping of absolute paths to project definitions: {"<path>": {"name": "memd", "extra_sources": ["*.jsonl"]}}.
global_root	string	HOME	Path to the global fallback project root.
global_brief_chars	integer	800	Max character excerpt of global state.md injected into a project's brief.

Default File Budgets

"budgets": {
  "state.md": 10000,
  "decisions.md": 12000,
  "todo.md": 10000,
  "mistakes.md": 22000
}

---

Built-In Credential Redaction

Session transcripts have a bad habit of containing API keys. Before anything reaches the curator model, memd scrubs it with a set of regex filters, so a token that leaked into a transcript doesn't leak again into a prompt (or into a git-committed memory file).

13 rules are built in:

• google_oauth (Google OAuth2 tokens, ya29. prefix)
• github_pat (Classic and fine-grained GitHub tokens)
• anthropic_key (Anthropic API keys, sk-ant- prefix)
• openai_key (OpenAI API keys, sk- / sk-proj- prefix)
• aws_access (AWS Access Key IDs, AKIA prefix)
• slack_token (Slack bot/user/workspace tokens, xox prefix)
• gitlab_token (GitLab PATs, glpat- prefix)
• npm_token (npm access tokens, npm_ prefix)
• jwt (JSON Web Tokens starting with eyJ)
• json_token_field (JSON keys containing access_token, refresh_token, etc.)
• bearer_header (HTTP Authorization headers with Bearer tokens)
• env_credential (UPPERCASE credentials assigned in .env or shell exports)
• ssh_private_key (PEM blocks starting with -----BEGIN ... PRIVATE KEY-----)

Note

Azure storage/service keys do not have a distinct prefix and cannot be reliably matched without high false-positive rates. Use REDACT_EXTRA_PATTERNS to configure rules matching your specific Azure keys if needed.

---

CLI Command Quick Reference

Command	Usage	Exit Codes & Notes
memd init [path]	Scaffold .memory/ & register a project. Supports --name <name> and --global (sets up the fallback root).	0 (Success), 2 (Config error)
memd sync	Distill new session content into memory. Supports --project <path>, --transcript <path>, --trigger <name>, and --dry-run.	0 (Success/No-op), 3 (Curator/distill failure)
memd sweep	Periodic sweep to catch up all projects, ingest inbox files, and detect new git projects. Supports --jobs <N>.	0 (Success), 1 (Any worker failed)
memd brief [path]	Print the session-start brief (context injection). Supports --max-chars <N> and --topic <keyword>.	Prints brief text to stdout.
memd status	Display registry, backlog bytes, and last distill summaries.	Prints status details.
memd install-hooks	Idempotently wire hooks into ~/.claude/settings.json.	Configures hooks.
memd note	Append a collision-safe note to the project or global inbox. Supports -m "<message>" and --global.	0 (Success), 2 (Config error)
memd exclude <path>	Exclude a path from automatic project discovery.	Registers path to exclude list.
memd hook <event>	Invoked by claude-code CLI lifecycle events.	session-start, session-end, pre-compact

---

Installation

pip Installation

Requires Python 3.10+. memd has zero runtime dependencies (standard library only).

# From PyPI
pip install memd

# With pipx (recommended for CLI tools — isolated, no system-Python friction)
pipx install memd

# Directly from the repository, no PyPI needed
pip install git+https://github.com/lowcache/memd

# From a local clone (add -e for editable/development mode)
pip install .

PEP 668 note: distributions that mark the system Python as externally managed (NixOS, Debian 12+, Arch, Fedora 38+) reject bare pip install outside a virtual environment. Use pipx, or a venv: python3 -m venv ~/.venvs/memd && ~/.venvs/memd/bin/pip install memd.

Nix Installation

memd is packaged as a Nix flake. You can run or install it directly:

# Run ad-hoc from the flake
nix run github:lowcache/memd -- status

# Install to user profile
nix profile install github:lowcache/memd

Declarative Setup via Home Manager

Import the module and enable the periodic sweep timer:

{ inputs, pkgs, ... }: {
  imports = [ inputs.memd.homeManagerModules.default ];

  services.memd = {
    enable = true;
    installClaudeHooks = true; # Idempotently configure ~/.claude/settings.json hooks

    sweep = {
      enable = true;            # Periodically execute `memd sweep`
      interval = "30min";       # Timer interval (default: "30min")
      onBoot = "5min";          # Delay after boot before running the first sweep (default: "5min")
      randomizedDelay = "2min"; # Randomized delay jitter (default: "2min")
    };
  };
}

Standalone systemd User Service & Timer

For non-Nix environments, a standalone systemd user service and timer can be set up using files in the contrib/ directory:

# Copy units to the user systemd directory
cp contrib/memd-sweep.* ~/.config/systemd/user/

# Reload the systemd user manager
systemctl --user daemon-reload

# Enable and start the timer
systemctl --user enable --now memd-sweep.timer

Refer to contrib/README.md for detailed customization and uninstallation instructions.

Manual Installation (no pip, no Nix)

memd is compatible with Python 3.10 to 3.13 and uses standard library packages exclusively, so a plain clone runs as-is:

git clone https://github.com/lowcache/memd && cd memd
./memd.py --help

Symlink the shim onto your PATH for a live-updating install: ln -s "$PWD/memd.py" ~/.local/bin/memd.

---

Testing & Verification

161 automated tests cover the parts that would hurt if they broke: XDG isolation, cursors, redaction, database parsing, atomic inbox publishing (including an 8-writer concurrency stress test), retry policies, brief budgeting, and curation quality. Verified on Python 3.10 through 3.13.

To run tests in a Nix-isolated environment:

nix flake check

Or run via pytest locally:

python -m pytest tests/ -q