grill-me

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security threats were identified. The skill implements a natural language workflow for interviewing users about their plans. It does not perform network requests, execute arbitrary shell commands, or include obfuscated content.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze user-provided plans or design documents (Step 0). While this represents a data ingestion surface, the skill's capabilities are limited to natural language interaction and read-only codebase exploration. There are no explicit boundary markers or sanitization steps defined, but the risk is negligible given the conversational nature of the tool.
  • [METADATA_POISONING]: The metadata in README.md and the SKILL.md frontmatter correctly describe the skill's purpose and author. No deceptive instructions or hidden payloads were found in the metadata fields.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 02:47 PM
Security Audit — agent-trust-hub — grill-me