Privacy story

Your data stays on your machine.

A plain-language explanation of where your data goes, and doesn't go, when you run TeamLaboration.

Data flow

~/.claude/projects/ (read-only)

Your existing Claude Code session files. TeamLaboration reads them; it never writes to them.

local daemon

A small background process running on your machine.

local SQLite database

Stores processed records (not raw logs). Lives at ~/Library/Application Support/TeamLaboration (mac) or %APPDATA%\TeamLaboration (Windows).

localhost:7842 dashboard

The web UI. Only reachable from your machine. No external network exposure.

claude CLI → Anthropic

The only outbound call. TeamLaboration passes individual conversation content to your local claude CLI, which talks to Anthropic on your own Claude subscription so the model can write the record. Nothing else leaves the machine.

What's stored

Processed records, structured 1–3 sentence descriptions per session.
Weekly rollups, once you generate them.
Configuration, output paths, polling interval, week boundaries.

What's not stored

· Raw Claude Code conversation logs (read on demand, not copied)
· Any API key, there is none; summaries run on your Claude subscription via the CLI
· Telemetry, usage analytics, error reports

No API key to handle

TeamLaboration never asks for or stores an API key. Summaries are produced by shelling out to your local claude CLI, which authenticates with your existing Claude subscription's own credentials. That means:

· No key is written to a config file or credential store by TeamLaboration
· Nothing is transmitted to a TeamLaboration server (there is no TeamLaboration server)
· Conversation content reaches Anthropic only through the CLI, when writing a record

No telemetry