grace-verification

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill identifies and executes verification tasks such as 'bun run gate:e2e' and 'grace lint' within the local environment to record software health evidence.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external configuration data from '.grace/verification/index.xml' which determines the parameters for command execution, creating a potential surface for indirect instruction injection.
  • Ingestion points: '.grace/verification/index.xml' and routed verification documents.
  • Boundary markers: Absent in instructions for parsing configuration files.
  • Capability inventory: Full shell command execution via defined gate tasks.
  • Sanitization: No mention of input sanitization or command validation for the verification manifest.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 03:49 PM
Security Audit — agent-trust-hub — grace-verification