icp-website-audit
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external websites during the audit phase, which creates a surface for instructions embedded in those sites to influence the agent's behavior.
- Ingestion points: The skill uses web crawling and fetch capabilities in Phase 2 (Website Scorecard Review) and Phase 3 (Competitive Head-to-Head) to retrieve content from user-provided company and competitor URLs.
- Boundary markers: The instructions do not specify the use of delimiters, XML tags, or explicit system instructions to treat the crawled website content as untrusted data rather than active instructions.
- Capability inventory: The skill has the capability to write multiple files to the local filesystem (e.g.,
icp-website-audit.md,personas.json, and various markdown reports) and invokes sub-skills such asbuyer-persona-generatorandcontent-asset-creator. - Sanitization: There is no mention of sanitizing, filtering, or escaping the content retrieved from external websites before it is used for synthesis and reporting.
Audit Metadata