security-app-multitenant
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's design involves ingesting and analyzing potentially untrusted data from an application's multi-tenancy layers, creating an attack surface for indirect prompt injection.\n
- Ingestion points: The skill processes data from various untrusted sources, including JWT claims, request headers, and subdomains (Section 4.1), application logs (Section 4.6), and background job payloads (Section 4.7) as described in SKILL.md.\n
- Boundary markers: The skill defines sensitive data redaction for its output, but it does not specify explicit boundary markers or instructions to disregard potential commands embedded within the external data (logs, database records) it is instructed to review.\n
- Capability inventory: The skill is granted access to powerful tools such as Bash, Write, and Edit in its frontmatter configuration in SKILL.md.\n
- Sanitization: The skill mandates the redaction of sensitive identifiers in the output, but there is no explicit sanitization process described for the data during its initial ingestion and processing phase.
Audit Metadata