openauth
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and mitigates indirect prompt injection risks by treating all authentication inputs as untrusted user data.
- Ingestion points: User-supplied inputs such as email addresses, OTP codes, and OAuth callback payloads (SKILL.md).
- Boundary markers: Includes the W011 guardrail which explicitly warns against treating inputs as instructions.
- Capability inventory: Server-side token issuance and data persistence to Cloudflare KV or DynamoDB.
- Sanitization: Recommends strict redirect URI allowlists and schema validation via Valibot or Zod.\n- [EXTERNAL_DOWNLOADS]: Fetches and uses legitimate authentication libraries from the official @openauthjs organization on NPM.\n- [CREDENTIALS_UNSAFE]: Enforces secure management of provider secrets (GitHub, Google, etc.) using environment variables and strictly forbids their exposure in client-side bundles.
Audit Metadata