grill-to-spec

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes git status --porcelain to monitor the workspace state. This is a read-only diagnostic command used to ensure no unintended changes occur during the documentation process.
  • [EXTERNAL_DOWNLOADS]: Recommends the user install auxiliary skills from a public third-party GitHub repository. The suggested command uses npm_config_ignore_scripts=true to mitigate the risk of malicious lifecycle scripts during installation.
  • [INDIRECT_PROMPT_INJECTION]: Ingests user-provided design input and local formatting templates to generate Architecture Decision Records and glossary terms, creating a surface for potential indirect instruction influence.
  • Ingestion points: User-provided design details and local format files (CONTEXT-FORMAT.md, ADR-FORMAT.md).
  • Boundary markers: None identified.
  • Capability inventory: Executes git commands and reads local file content.
  • Sanitization: No evidence of input validation or escaping for user-supplied content used in documentation drafts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 09:35 AM
Security Audit — agent-trust-hub — grill-to-spec