tanner-cleanup

Pass

Audited by Gen Agent Trust Hub on Jul 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted data from git diff output (newly added lines in comments, strings, and docs), creating an attack surface for indirect prompt injection. * Ingestion points: git diff output from the current repository. * Boundary markers: Absent. * Capability inventory: Shell command execution (linters) and file system writes (state fingerprinting). * Sanitization: Absent.
  • [COMMAND_EXECUTION]: The skill executes various shell commands to identify changed files, run repository-specific linters, and manage commit state. * Evidence: Includes commands like 'git diff --cached --name-only', 'npm run lint', and 'git write-tree'. * Context: The skill writes a state file to the .git directory ('tanner-cleanup.ok') to signal completion to pre-commit hooks; while writing to the .git directory is a sensitive operation, it is used here for legitimate state management.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 9, 2026, 05:01 PM
Security Audit — agent-trust-hub — tanner-cleanup