memento-brain-collector 0.0.3

Cross-platform agent that collects AI coding tool memory files and syncs to a central server

memento-collector

Cross-platform agent that automatically collects AI coding tool memory files and syncs them to a central Memento server (a shared brain for your AI coding tools).

Supported AI Tools

Tool	Data Collected
Claude Code	Conversations, memory, plans, history
OpenClaw	Sessions, identity, memory, learnings, skills
Codex	Sessions (active + archived), history, skills
Antigravity	Full conversations (built-in .pb decryption — AES-256-GCM + protobuf), brain plans, code snapshots
Obsidian	All markdown notes in your vault
Cursor	Agent transcripts, skills, MCP config

Install

pip install memento-brain-collector

Antigravity support (decrypting encrypted .pb conversation files) is built in — no extras needed. The cryptography library is already a required dependency.

Quick Start

# Interactive setup (first time)
memento-collector setup

# Or run directly
memento-collector run

The setup wizard will:

1. Detect your platform (macOS / Linux / Windows)
2. Auto-discover installed AI tools and Obsidian vaults
3. Configure the server URL and auth token
4. Optionally install as a system service (auto-start on boot)

Commands

memento-collector setup      # Interactive setup wizard
memento-collector run        # Run in foreground
memento-collector install    # Install as system service
memento-collector start      # Start the service
memento-collector stop       # Stop the service
memento-collector status     # Show collector status
memento-collector uninstall  # Remove system service

How It Works

1. File Watching — Uses watchdog (FSEvents on macOS, inotify on Linux, ReadDirectoryChanges on Windows) to detect file changes in real-time
2. Parsing — Supports Markdown, JSONL, JSON, TOML, SQLite formats
3. Sanitization — Automatically redacts API keys, tokens, passwords, private keys before upload
4. Queuing — Local SQLite queue for offline resilience (syncs when server is reachable)
5. Syncing — HTTP upload to server, with chunked upload for files > 2MB (tested with 37MB files)
6. Device Identity — Each device gets a persistent unique ID, all data tagged with device info

Configuration

Environment variables (or set via memento-collector setup):

Variable	Default	Description
MEMENTO_SERVER_URL	http://localhost:8001	Server API URL
MEMENTO_SERVER_TOKEN		Collector auth token
MEMENTO_OBSIDIAN_VAULT_PATH	Auto-detected	Obsidian vault path

Config file: ~/.memento/config.json

System Service

Platform	Service Type	Config Location
macOS	LaunchAgent	~/Library/LaunchAgents/com.memento.collector.plist
Linux	systemd user	~/.config/systemd/user/memento-collector.service
Windows	Task Scheduler	MementoCollector scheduled task

License

MIT