audit-website
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires downloading the squirrel CLI from the vendor's official website (squirrelscan.com) and includes commands for self-updates and additional skill installations.
- [REMOTE_CODE_EXECUTION]: The tool includes self-update and skill installation features (
squirrel self update,squirrel skills install) that download and execute code from the vendor's infrastructure. - [COMMAND_EXECUTION]: The skill uses the Bash tool to execute the
squirrelbinary to perform crawls, analysis, and report generation. - [PROMPT_INJECTION]: The skill presents an Indirect Prompt Injection surface.
- Ingestion points: Content from external websites is ingested via
squirrel auditandsquirrel reportcommands (SKILL.md). - Boundary markers: No instructions are provided to the agent to distinguish between tool analysis and potentially malicious instructions embedded in the audited website content.
- Capability inventory: The agent is granted
Read,Edit,Grep, andGlobtools and is explicitly encouraged to automate code and content fixes based on the audit reports. - Sanitization: There is no evidence of sanitization for the data retrieved from external websites, creating a path for malicious instructions on those sites to influence the agent's local code modifications.
Audit Metadata