verification-first

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local commands including pnpm for linting, testing, and building, as well as a local PowerShell script .\scripts\health-check.ps1 to verify project health. These operations are standard for the skill's stated purpose of code verification.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it requires the agent to ingest and report tool outputs and error messages verbatim. This creates a surface where malicious instructions embedded in error messages or logs could influence agent behavior.
  • Ingestion points: Command output and verbatim error messages mentioned in SKILL.md.
  • Boundary markers: None specified in the verification report format.
  • Capability inventory: Execution of local scripts and shell commands (pnpm, .ps1) defined in SKILL.md.
  • Sanitization: No sanitization or filtering of tool output is prescribed.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 11:48 AM