branch-surgery-pr-split

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from the local git environment, including commit messages, branch names, and diff outputs.\n
  • Ingestion points: Git commit counts, churn data, and file lists are gathered in SKILL.md (Step 1) and references/audit-gates.md.\n
  • Boundary markers: The instructions do not define technical delimiters for wrapping git output, but they mandate multiple user collaboration and approval steps.\n
  • Capability inventory: The skill utilizes shell-based git commands such as diff, log, and cherry-pick to manipulate repository state.\n
  • Sanitization: Verification is performed through 'audit gates' in references/audit-gates.md, which require strict parity checks between the original and split branches.\n- [COMMAND_EXECUTION]: The skill uses standard git command-line tools to analyze and restructure branches. These operations are essential to the primary function of the skill and are performed within the local repository context.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 02:36 PM