agency-quick
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection via external web content.
- Ingestion points: The
WebFetchcommand inSKILL.mdingests untrusted data from arbitrary URLs provided by the user. - Boundary markers: The skill does not implement boundary markers or specific instructions (e.g., "treat the following as data only") to prevent the agent from executing instructions found on the target webpage.
- Capability inventory: The agent's capabilities are limited to data analysis and terminal output; no high-risk capabilities like file writing or subprocess execution are present.
- Sanitization: There is no evidence of content filtering or sanitization applied to the retrieved webpage data before it is processed by the LLM.
Audit Metadata