review-spec
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted user-controlled documents to generate findings that are written to the filesystem.
- Ingestion points: The skill reads
spec.md,delta/files,log.md, andseed.md(as specified inSKILL.md) to assess planning readiness. - Boundary markers: The instructions do not define explicit delimiters or instructions to the model to ignore potential commands embedded within these input documents.
- Capability inventory: The skill performs file-write operations to the
.pending-reviews/directory using theemit-pending-review.mdtemplate when findings are identified. - Sanitization: There is no evidence of content sanitization or escaping for user-controlled text (like quotes from the spec) before it is recorded in the output findings bundle.
Audit Metadata