Security & privacy
ForgeDeck's security model — local-first core, redaction, and cloud boundaries.
Embed
<a href="https://docs.forgedeck.net/security"><img src="https://docs.forgedeck.net/_api/public/orgs/db9cb8ff-b606-46cf-bbcd-02c8f8a9a3e2/spaces/6f417400-93ca-41a8-a914-df2ae3981dc6/pages/addc7b9e-8ef9-4d61-98fc-eb1cbec91c18/social-card.svg?v=3e7ae33eec8b" alt="Security & privacy" width="600" height="315"></a>Local-first by default
The desktop core runs entirely on your machine. Your source code, secrets, and project data never leave your device for the core flow — terminals, Kanban, worktrees, review, and local memory are all local.
The process boundary is the security boundary
Dangerous capabilities — PTYs, shell exec, SQLite, filesystem writes, Git, MCP server — live only in the Rust backend. The renderer has no direct OS access.
Secret redaction
Secrets are redacted before any memory write or prompt attachment — API keys, tokens, and .env values never reach .forgememory or a prompt bundle.
Opt-in, confirmed danger
Destructive MCP/Git actions require both a settings opt-in and a UI confirmation. The local MCP server is disabled by default.
What the cloud processes
If you create an account, the cloud handles auth, billing (Stripe — no card numbers stored), licensing/entitlement checks, and any server-executed feature you invoke. The desktop also sends anonymous usage & diagnostic telemetry and phones home to validate your license. It never sends your code. Session tokens are stored hashed; passwords use scrypt. See the Privacy Policy.