growth-strategy
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill collects data from external sources to inform its growth strategy recommendations.
- Ingestion points: Researches company websites and public sources to identify product positioning and competitors. It also reads the .agents/product-context.md file and ingests user-supplied CSV data or pasted figures.
- Boundary markers: The references/house-rules.md file contains a security rule (Rule 7) instructing the agent to treat fetched or pasted content as data rather than instructions. However, the skill does not implement specific delimiters for the external content in its primary logic.
- Capability inventory: The skill writes researched and inferred data back to the .agents/product-context.md file and generates prioritized growth plans.
- Sanitization: No explicit data sanitization or validation routines for external content were identified.
- [PROMPT_INJECTION]: The skill contains instructions to suppress certain types of output to the user.
- Evidence: In SKILL.md, the agent is instructed: 'Do not return a draft with a warning on it. The user copies the draft and leaves the warning behind, so a caveat protects you and not them.' It is further instructed to 'Ask only for what research genuinely cannot establish' to maintain a three-question limit.
- Context: These instructions are designed to improve user experience and ensure the agent gathers necessary data rather than providing potentially flawed drafts. While not malicious, this is a form of output control and concealment of operational warnings.
Audit Metadata