general-tasks
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface because it defines a process for handling tasks involving external files and web content without implementing security boundaries.
- Ingestion points: Metadata description specifies tasks involving 'files' and 'the web'.
- Boundary markers: Not present. The instructions lack delimiters or warnings to ignore data-embedded instructions.
- Capability inventory: The skill facilitates operations involving shell execution, file modification, and web access.
- Sanitization: Not present. No validation or filtering of external inputs is defined.
Audit Metadata