code-polish

Pass

Audited by Gen Agent Trust Hub on Oct 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes git diff output and local file contents as primary inputs for its cleanup tasks. These inputs could contain malicious instructions embedded in comments or code designed to influence the agent's behavior.
  • Ingestion points: Output from git diff, git status, and git ls-files (SKILL.md, Step 0).
  • Boundary markers: None defined; the agent processes the raw diff content.
  • Capability inventory: Executes shell commands including git, grep, cat, and project-defined build, lint, and test scripts (e.g., npm test, pytest, cargo test).
  • Sanitization: No explicit sanitization or filtering of the diff content is performed before the agent analyzes it.
  • [COMMAND_EXECUTION]: The skill relies on executing various shell commands to inspect the repository state and verify the results of the code polish.
  • Specifically, it uses git for diffing, grep for searching patterns, and standard ecosystem tools like npm, npx, ruff, mypy, pytest, cargo, and go for formatting, linting, and testing.
  • These commands are standard for development workflows and are used here within their expected scope.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 4, 2026, 12:04 AM