grill-with-docs
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns such as prompt injection, obfuscation, or persistence mechanisms were detected. The skill uses configurations like
disable-model-invocation: trueandallow_implicit_invocation: falseto restrict its execution environment and reduce the attack surface.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes user-provided plans and designs to generate project documentation.\n - Ingestion points: User input provided during the design interview process described in
SKILL.md.\n - Boundary markers: None explicitly defined to separate untrusted user content from instructions.\n
- Capability inventory: Limited to writing markdown content to
CONTEXT.mdand thedocs/adr/directory. No shell execution, network exfiltration, or sensitive file access capabilities are present.\n - Sanitization: Not present.\n
- Context: This represents the primary functional surface for a documentation skill and is evaluated as safe given the lack of exploitable tools.
Audit Metadata