secrets-exposure-review
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has a defined surface for indirect prompt injection as it is designed to ingest and analyze untrusted external data such as source code, CI logs, and configuration files.
- Ingestion points: Untrusted data enters the agent context via files described as 'supplied authorized code/config/CI/log/artifact/history' as specified in
prompts/secrets-exposure-review.md. - Boundary markers: The instructions in
prompts/secrets-exposure-review.mdinclude explicit directives to separate facts from inferences and to distinguish between placeholders and real candidates, providing a logical boundary for processing. - Capability inventory: The skill is primarily analytical.
evals/local-rules.jsonrestricts the agent's environment by settingmax_escalationsto 0, and the instructions strictly prohibit calling real APIs or performing destructive actions. - Sanitization: The core domain procedure mandates the use of redacted fingerprints and locations, explicitly forbidding the copying of values or reading of real credentials.
Audit Metadata