update-analogy-catalog
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads content from newly published posts to extract analogies and metadata for a shared catalog. This creates a surface for indirect prompt injection if a post contains instructions designed to manipulate the catalog update process.
- Ingestion points: Reads the published post file (e.g.,
corpus/published/{section}/). - Boundary markers: The skill lacks explicit instructions to ignore embedded commands or instructions within the post content being analyzed.
- Capability inventory: Performs file-read operations on posts and file-append operations to
shared-context/analogy-catalog.md. - Sanitization: No explicit sanitization or validation of the extracted analogy strings or 'why it worked' notes is mentioned before appending to the markdown table.
Audit Metadata