preflight-regression

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes Git commands and optionally runs project-specific npm scripts (test, build, lint, typecheck). These are standard development tasks performed in the user's local repository context. The check-runner.mjs script specifically uses spawn with shell: false and the --ignore-scripts flag to prevent unauthorized lifecycle hook execution (pre/post scripts). It also implements a timeout and output limit to mitigate resource exhaustion or accidental large-scale logging.
  • [DATA_EXFILTRATION]: The skill identifies sensitive files (e.g., .env, id_rsa, .pem) using the sensitive function in run-preflight.mjs and explicitly skips reading their content to prevent accidental leakage into the AI agent's context. Findings regarding these files only report the path without disclosing values.
  • [EXTERNAL_DOWNLOADS]: The skill does not perform external downloads or network requests. It validates local relative links in documentation but ignores external URLs, ensuring operation remains local to the repository.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns were detected. The skill only executes local binaries (Git, Node) and local project scripts that are explicitly authorized by the user via CLI arguments.
  • [PRIVILEGE_ESCALATION]: The skill runs with the user's existing permissions and does not attempt to use sudo or modify system-level configurations. It uses taskkill.exe on Windows to clean up timed-out subprocesses, which is a standard procedure for local script management.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 07:01 AM