grilling
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill automatically retrieves information from the environment (filesystem, tools, etc.) to answer its own questions, which creates a vulnerability surface for data-driven prompt injection.
- Ingestion points: Data is ingested from the filesystem and tool outputs to find 'facts' in SKILL.md.
- Boundary markers: The instructions do not include boundary markers or delimiters to isolate untrusted data from system instructions.
- Capability inventory: The skill dispatches sub-agents with access to filesystem and command-line tools. Although sub-agents use a 'fresh context window', the resulting facts are returned to the main context.
- Sanitization: There are no instructions to sanitize or escape the content retrieved from the environment before it enters the agent's context.
Audit Metadata