Privacy policy.

What we collect

Anonymous usage telemetry (on by default)

To understand how Foundry is being used and where it breaks, the app sends a small set of anonymous events:

• session_start and session_end when you open and close the app
• wizard_complete with the mode you chose, the number of agents generated, and the number of capabilities installed
• forge_agent with the agent role and kind when you create an agent in the Forge editor
• refinement_approved with the agent name when you accept a refinement proposal
• blueprint_import and blueprint_export with the number of agents involved

Every event includes the app version, your operating system, the number of beta days remaining, and a random session ID that is not persisted between launches.

Bug reports and feedback (only when you send them)

If you click the feedback button inside the app, we receive the category you picked and the message you wrote. Nothing more.

What we never collect

• Your source code, file contents, or file paths
• Your project names or repository URLs
• Your Anthropic API key or any other credentials
• Your agent instructions, hook scripts, or rule files
• Anything you type into the wizard interview or chat with Scholar
• Personal identifiers: your name, email, IP address, machine ID, or browser fingerprint

What stays on your machine

Foundry does most of its work locally:

• Graphify builds the semantic index of your codebase entirely on your computer. No tokens spent, no data uploaded.
• MemPalace stores its memory graph in a local SQLite + vector database under your project's .mempalace/ directory. Each project's memory is yours alone.
• Generated agents, hooks, rules, and standards are written to your project's .claude/ directory as plain files.
• Settings, API keys, and telemetry preferences live in your OS user directory and are encrypted via your operating system's keychain when sensitive.

Third-party services

Anthropic (Claude API)

Foundry is bring-your-own-key. When Scholar runs an interview or a generation step, your project context and your interview answers are sent directly from your machine to Anthropic's API using your API key. That data is governed by Anthropic's terms of service and privacy policy. Foundry does not see, log, or proxy this traffic.

GitHub (auto-updates)

Foundry checks for new versions by fetching the public release manifest from GitHub. The request is anonymous; no account or identifying data is sent.

Google Sheets webhook (telemetry storage)

Anonymous telemetry events and feedback submissions are POSTed to a Google Apps Script webhook that writes them into a Google Sheet we maintain. The sheet is private to the Foundry team. Standard Google network metadata applies (IP address visible to Google during the request). We do not store or process IP addresses on our end.

Data retention

Telemetry events and feedback submissions are kept indefinitely in the Google Sheet so we can study product trends over time. There is no scheduled purge. If you want a specific entry deleted, email [email protected] with a description of the entry and we will remove it.

Opting out of telemetry

Open Foundry, go to Settings → Privacy, and turn off telemetry. Once disabled, no events are sent. The first-run notice in the app explains this and links to the same setting. Feedback submissions remain a one-time, user-initiated action — turning off telemetry does not affect your ability to send a bug report when you want to.

Children

Foundry is a developer tool and is not directed at children under 13. We do not knowingly collect data from anyone under 13.

Changes to this policy

If this policy changes materially, we will update this page and bump the date at the top. Past versions are visible in the public commit history of try-foundry/tryfoundry-site.

Contact

Questions, deletion requests, or anything else: [email protected].