Skip to content

SECURITY.md

Latest commit

 

History

History
30 lines (23 loc) · 1.4 KB

SECURITY.md

File metadata and controls

30 lines (23 loc) · 1.4 KB
Raw

Security Policy

Reporting a vulnerability

Please report security issues privately via GitHub's "Report a vulnerability" (Security advisories) rather than opening a public issue. Include a description, reproduction steps, and impact. We aim to acknowledge within 72 hours.

Threat model and handling of secrets

ownllm stores OAuth credentials for your personal LLM subscriptions. They are sensitive bearer secrets.

  • At rest: the auth store (~/.ownllm/auth.json) is written with mode 0600 inside a 0700 parent directory, via atomic create-and-rename.
  • In logs: credentials are never logged. Logging redacts authorization, cookie, set-cookie, *.access_token, *.refresh_token, *.id_token, and cf_clearance. Request/response bodies are logged only at debug.
  • Client authentication: API requests require server.api_key, compared in constant time. The server refuses to start on a non-loopback host without one.
  • Network egress: the upstream HTTP client is host-pinned — bearer tokens and cookies are only ever sent to the allowlisted upstream hosts for each provider.
  • No telemetry: ownllm makes no outbound calls other than to provider OAuth and inference endpoints.

Scope and intended use

ownllm is for personal, single-user use of your own subscriptions. See the disclaimer in the README. It is not designed or supported for multi-tenant or resale scenarios.