commit-fixup

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill relies on executing various Git commands (git log, git rebase, git commit, git apply) to automate history rewriting. These commands are necessary for the skill's primary function and are implemented with safety constraints, such as aborting rebase on conflict and creating backup branches.
  • [PROMPT_INJECTION]: The skill processes untrusted repository data, including Git commit logs, status outputs, and diff/patch content. This presents a surface for indirect prompt injection where malicious content in a repository could attempt to influence the agent's logic.
  • Ingestion points: Repository data via git log, git status, git diff, and patch files in SKILL.md and references/edge-cases.md.
  • Boundary markers: The skill explicitly requires displaying a mapping table to the user and obtaining confirmation via AskUserQuestion before execution.
  • Capability inventory: The agent has the capability to execute shell commands (git) and write to the local file system (e.g., /tmp/ patches).
  • Sanitization: There is no explicit sanitization of Git log entries or patch content, but the verification phase (diffing against a backup branch) provides a post-execution safety check.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 01:12 PM
Security Audit — agent-trust-hub — commit-fixup