We support the latest stable release. Older releases receive fixes on a best-effort basis.
Please report security vulnerabilities privately via GitHub's "Report a vulnerability" flow on the Security tab:
https://github.com/endara-ai/endara-relay/security/advisories/new
Please do not open public issues or pull requests for security reports.
- We aim to acknowledge new reports within 3 business days.
- We aim to provide a status update within 7 business days of acknowledgement.
- We will coordinate disclosure timing with the reporter and credit you (with permission) in the resulting advisory.
In scope
- The
endara-relaybinary, its OAuth callback handler, and the management API (Unix-domain socket on macOS/Linux, Named Pipe on Windows). - The
endara-desktopTauri app shell and the IPC commands it exposes (reported in the desktop repo, but accepted here too if it's unclear which side is affected).
Out of scope
- Third-party MCP servers configured as upstreams. The relay faithfully forwards their responses; sandboxing upstreams is a non-goal (see
THREAT_MODEL.md→ "Known residual risks"). - User-misconfigured token directories (e.g. tokens placed on cloud-synced paths like Dropbox or iCloud Drive). The app surfaces an in-product warning but cannot prevent it.
- Issues only reproducible by an attacker who already has administrative or local code-execution access as the same OS user.
See THREAT_MODEL.md for the full threat model and trust boundaries.