Data Storage and Artifacts

Metateam is local-first. It stores configuration, state, and project artifacts on your filesystem. Understanding where these files live is essential for backup, cleanup, and debugging.

User Home Directory (`~/.metateam/`)

This directory contains global state and configuration for your user account.

File/Path	Purpose
`config.json`	Global settings and your Metateam API key.
`crew.db`	SQLite database storing crew operational state and local message history.
`log/unified.log`	Central log file for all Metateam CLI and hook activity.
`repos/`	Directory containing local FTS5 code search index databases (one `.db` per repo).

Project Root Artifacts

When you use Crew or Attractor, Metateam creates directories in your project root to store durable artifacts.

Crew Artifacts (`.crew/`)

File/Path	Purpose
`history.md`	A human-readable audit trail of all crew activity in this project.
`mail-board/`	A directory of markdown reports and long-form communications between agents.
`<persona>/`	(Internal) Temporary working directories for individual agents.

Attractor Artifacts (`.attractor/`)

File/Path	Purpose
`runs/`	Chronological history of pipeline runs. Each run has its own subdirectory with stage status and artifacts.

Hidden Markers

Metateam uses small marker files to control behavior without global configuration.

File	Purpose
`.metateam.json`	Pins a directory tree to a specific Metateam project.
`.nometateam`	Disables all Metateam hooks and monitoring for this directory and its children.
`.yesmetateam`	Overrides a `.nometateam` marker in a parent directory to re-enable Metateam in a specific subtree.

Backup and Migration

API Key: If you move to a new machine, your API key in config.json is the only thing you must migrate to access your hosted session history and knowledge base.
Local Databases: The SQLite databases (crew.db and repo search .db files) can be safely deleted; they will be recreated as needed, though local message history and search indexes will be lost.
Project Artifacts: Treat .crew/ and .attractor/ like your .git/ directory. They contain the "memory" and audit trail of your project’s AI collaboration.