detect-product-mismatches
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze content from external, potentially untrusted sources, which creates a surface for indirect prompt injection where malicious instructions could be embedded in data to influence the agent.
- Ingestion points: The skill reads data from source code, issue trackers (Jira, GitHub), requirements documents (Confluence), and public documentation sites (documented in SKILL.md and references/features.md).
- Boundary markers: The instructions do not define specific delimiters or security boundaries to isolate external content from the agent's internal reasoning logic.
- Capability inventory: The skill utilizes command-line tools like
ghandgitfor data retrieval and has write access to the local filesystem to generate reports in themismatches/directory (documented in SKILL.md steps 1 and 5). - Sanitization: Although references/features.md instructs the agent to "restate, never paste," this is a formatting guideline rather than a security filter to prevent the execution of instructions contained within the ingested text.
Audit Metadata