git-commit-rewrite

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill automates complex Git workflows using shell commands such as git filter-branch, git push --force-with-lease, and git cherry-pick. These are necessary for the primary function of rewriting repository history.
  • [COMMAND_EXECUTION]: It executes a local Python utility script (scripts/rewrite_msg.py) to apply message transformations. This execution is limited to the skill's own packaged code.
  • [PROMPT_INJECTION]: The skill processes potentially untrusted data in the form of existing Git commit messages (indirect prompt injection surface).
  • Ingestion points: Recent Git commit subjects and bodies (SKILL.md, step 4).
  • Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the messages being processed.
  • Capability inventory: The skill has significant capabilities including file system modification (/tmp/cc-rewrite-map.tsv), shell command execution, and network communication via Git push.
  • Sanitization: No specific sanitization of commit message content is described; however, the workflow mandates that the agent 'Show old -> new plan and wait for explicit confirmation' (step 6), providing a human-in-the-loop defense against malicious content influencing the final rewrite.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 01:12 PM