write-product-brief
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (user feedback and support cases) to generate product briefs, which introduces a potential vector for malicious instructions to influence the agent's behavior during the brief creation process.\n
- Ingestion points: The 'Gather Product Evidence' section in
SKILL.mdidentifies 'user feedback or support cases' as external signal sources that the agent must collect.\n - Boundary markers: Absent; the instructions do not specify the use of delimiters or instructions to ignore embedded commands within the gathered feedback.\n
- Capability inventory: The skill's configuration in
SKILL.mdpermits access to several sensitive tools, includingwrite_file,edit_file,create_file, andexecute_command.\n - Sanitization: Absent; although the skill includes logical guardrails to cross-check goals, it lacks technical instructions for escaping, filtering, or validating the content of the external feedback strings.
Audit Metadata