What we collect, and what we don't.

The website and docs

We measure basic traffic so we know what is useful, using privacy-first analytics:

• The marketing site (www.claw.boo) uses Cloudflare Web Analytics for page views and performance, and PostHog (US-hosted) in cookieless mode to measure which calls-to-action and features people use. PostHog's autocapture is enabled, so it records interaction events across the site (clicks, form submissions, and the text and attributes of the element you interacted with) alongside the page URL. Neither sets cookies nor stores anything in your browser, and PostHog does not identify you across sessions. Both process your IP address to serve and count requests, acting as our processors. If your browser sends a Do Not Track signal, we skip loading PostHog altogether, so no analytics request is made at all.
• The documentation (docs.claw.boo) runs on Mintlify, which provides built-in analytics. It sets no cookies, but it stores an anonymous identifier in your browser's local storage to count unique visitors, and processes standard request data such as your IP address and browser server-side.

We do not set advertising cookies, run ad pixels, do cross-site tracking, record your session, or sell data.

The Clawboo software

Clawboo is local-first. Your board, chat history, memory and settings stay on your own machine: the board lives in SQLite at ~/.clawboo/clawboo.db, and your API keys sit in an AES-256-GCM encrypted vault at ~/.clawboo/secrets/. There is no Clawboo account and no product usage tracking.

Out of the box, Clawboo makes only the network requests listed here. Your prompts and code go to the model provider you configure. On startup and on demand it asks the npm registry for the latest published version, so it can tell you an update exists. It fetches the marketplace catalogue of agents and teams from this project’s own GitHub repository, cached for six hours. If you have configured a provider key, it asks that provider which models your key can use. To link related memories, it indexes each fact with a local Ollama when one is running with its embedding model, which keeps the text on your machine; with no such Ollama and an OpenAI key available to Clawboo (one you connected, or an OPENAI_API_KEY in its environment), each fact’s text, and the text of memory searches (including the recall an agent run makes when it starts), is sent to OpenAI instead, and Memory says so. Once your memory has been indexed locally, it never moves to OpenAI unless you choose that. The choice lasts until Clawboo next finds your local Ollama serving, or until you disconnect the key (a key exported in Clawboo’s environment stays available until you unset it), and Memory shows while it is in effect. The dashboard loads its typeface from Google Fonts. An OpenTelemetry trace bridge is available and is off until you enable it and point it at a collector you control.

Anything you switch on reaches what you point it at. A connector or MCP server talks to its own host, an agent given a browser reaches whatever it browses, and installing a connector on first use pulls its package from npm or PyPI. Those are your choices to make, and they are not covered by the list above.

If we add optional product telemetry in the future, we will document exactly what it collects, keep it privacy-respecting, and update this page. The source is on GitHub.

Contact

Questions about privacy: conduct@claw.boo.