search-specialist
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to search for and extract content from external websites, which introduces a surface for indirect prompt injection where malicious instructions on a webpage could influence the agent.
- Ingestion points: SKILL.md (under 'WebFetch Deep Dive') instructs the agent to extract full content from external web pages and follow citation trails.
- Boundary markers: The instructions do not provide specific markers or guidelines to distinguish fetched content from system instructions, nor do they tell the agent to ignore instructions embedded in the external data.
- Capability inventory: The skill uses the agent's web browsing and synthesis capabilities. No file system access or command execution is specified in the skill content.
- Sanitization: No sanitization or validation of the ingested external content is mentioned or required by the instructions.
Audit Metadata