build-backlinks
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external platforms during the Channel Research phase, creating a surface for indirect instructions to influence the agent.
- Ingestion points: Web search results from Hacker News, Quora, GitHub, Wikipedia, and various industry forums are ingested by the agent during Phase 2 of the skill execution (
SKILL.md). - Boundary markers: The instructions do not define clear delimiters or "ignore embedded instructions" warnings for the content retrieved from web searches.
- Capability inventory: Across its scripts and instructions, the skill has the ability to read local brand documentation (
brand_dna.md), perform web searches, and write the final action plan to the local workspace (backlink_plan.md). - Sanitization: The skill does not implement specific sanitization or filtering of external content before the agent uses it to draft response templates or prioritize opportunities.
Audit Metadata