competitive-feature-benchmark
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from web searches which could contain malicious instructions.
- Ingestion points: Web search results are gathered in Step 2 of
SKILL.mdto analyze competitor features. - Boundary markers: The
Guardrailssection inSKILL.mdexplicitly instructs the agent to treat search results as untrusted data and ignore any embedded directives or instructions. - Capability inventory: The skill utilizes web search capabilities for research; no file-writing, command execution, or other high-privilege capabilities are defined in the procedure.
- Sanitization: Specific instructional guardrails are provided to ensure search content is treated strictly as observation material, preventing it from altering the skill's procedure or output structure.
Audit Metadata