validator

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill's primary function involves running shell commands based on scripts found in project files like package.json.
  • Evidence: Step 3 instructions to execute 'npm run lint', 'bundle exec rake test', etc., based on detected configuration files.
  • Risk: Malicious repositories can use these script definitions to execute harmful code on the user's machine when the validation skill is invoked.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill ingests data from local configuration files and command outputs to build a summary report.
  • Ingestion points: Local files (package.json, pyproject.toml, etc.) and tool stdout/stderr.
  • Boundary markers: Absent. The skill does not use specific delimiters to isolate untrusted content.
  • Capability inventory: Execution of shell commands (Step 3).
  • Sanitization: Absent. Output is parsed and inserted directly into the report markdown.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:15 PM