linkbuilding
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch external data from user-supplied domains and Google search results to perform phase assessments. This introduces a surface for indirect prompt injection where untrusted text on external homepages could attempt to influence the agent's execution flow.
- Ingestion points:
SKILL.mdunder Step 1 instructs the agent to fetch the homepage of the domain and perform Google queries for the brand name. - Boundary markers: Absent; no specific instructions or delimiter boundaries are provided to isolate or safely ignore potential instructions embedded within the external text.
- Capability inventory: No executable code or custom script files are distributed with this skill; it relies purely on content analysis and report text generation.
- Sanitization: Absent; web content is read and processed directly to extract visible signals without pre-filtering or parsing restrictions.
Audit Metadata