grill-with-docs
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and analyze untrusted data from the local codebase, creating a surface for indirect prompt injection.
- Ingestion points: The agent is directed to "explore the codebase" and search for existing documentation such as
CONTEXT.md,CONTEXT-MAP.md, and ADR files inSKILL.md. - Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat codebase content as untrusted or to ignore embedded instructions within those files.
- Capability inventory: The skill has the capability to modify existing files (
CONTEXT.md) and create new architectural decision records (ADRs) within thedocs/adr/directory. - Sanitization: The skill does not implement any sanitization, filtering, or validation of the content discovered during codebase exploration before incorporating it into the domain model or documentation.
Audit Metadata