FughettaSecurity
What protects your source
Fughetta stores source code, so the security question is a short one: who can reach it, and what can they do once they have. This page answers both without a marketing adjective.
- What reaches the gateway
- One MCP endpoint and one admin API, both behind a bearer token. A request for a workspace the token cannot reach fails as if the workspace did not exist, so other teams' names are not disclosed.
- Tokens
- Each software installation has its own token, scoped to a workspace at creation and given an expiry. Only its SHA-256 hash is stored; the raw token is shown once. Revoking it takes effect when the gateway's accounts snapshot refreshes.
- Your code
- Fughetta makes no model calls and runs no project code. Files and revisions live in the source; the gateway never writes them directly. Self-hosting is a supported deployment if you need the data in your own network.
- Transport
- HTTPS to the gateway and from the gateway to the source. A self-hosted gateway can bind to a LAN address for devices on your network; do not expose it beyond the network you trust.
- Isolation
- Each worldline is a separate edit history with undoing, redoing and snapshots, so one agent's changes never land in another's without a merge that reports its conflicts.
- Recoverability
- A change sent with an operation ID is recorded before it is answered, so a lost reply can be read back by its receipt instead of being retried.
Reporting a vulnerability
Write privately before writing anywhere else, and give us a chance to fix it. Include what you did, what happened, and what an attacker could do with it.
Write from the console: this deployment has not published an address.
Scope
Please do not test against workspaces you do not own, and do not run automated scanners against the hosted gateway. Both count as an attack on other people's source, and are handled as one.