multi-tenant
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill contains hardcoded default administrative credentials (admin/admin) within documentation examples for configuring Keycloak protocol mappers in a development environment. \n- [INDIRECT_PROMPT_INJECTION]: The skill implements patterns that ingest and process untrusted data from JWT claims and user-provided provisioning inputs, which are subsequently used in database operations and system configurations. \n
- Ingestion points: External data enters through JWT claims (org_id, org_name) and organization provisioning API request bodies (SKILL.md). \n
- Boundary markers: The provided patterns do not include explicit delimiters or safety instructions to prevent the agent from processing malicious commands embedded within tenant data. \n
- Capability inventory: The skill uses tools like Bash, Write, and WebFetch to perform database operations and administrative API calls to Keycloak (SKILL.md). \n
- Sanitization: Demonstrates basic regex-based sanitization for domain inputs but lacks comprehensive validation for other tenant-controlled fields before they are used in business logic.
Audit Metadata