geo-fix-llmstxt
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 content from external websites to generate site summaries and inventories, creating a vulnerability surface for indirect prompt injection.
- Ingestion points: The skill fetches data from user-supplied URLs, including the homepage, robots.txt, sitemaps, and up to 15 internal pages (SKILL.md, Phase 1).
- Boundary markers: The instructions explicitly direct the agent to wrap fetched content in
<untrusted-content>tags to isolate it from the main prompt context (SKILL.md, Security section). - Capability inventory: The agent uses network fetching to retrieve web content and file writing to save
llms.txtandllms-full.txtto the local workspace. - Sanitization: The skill contains specific instructions to ignore common injection patterns like 'Ignore previous instructions' or 'You are now...' if they appear within the crawled content (SKILL.md, Security section).
Audit Metadata