split-to-prs
Pass
Audited by Gen Agent Trust Hub on Jul 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted external data from repository diffs and history, creating a surface for indirect prompt injection.
- Ingestion points: Processes git diffs, uncommitted changes, chat history, and ownership files like
CODEOWNERS(SKILL.md). - Boundary markers: No explicit delimiter markers are used for external data; however, the skill mandates a human-in-the-loop 'propose and approve' workflow.
- Capability inventory: Executes git commands including branch creation, staging, committing, and pushing (SKILL.md).
- Sanitization: The skill explicitly avoids dangerous bulk commands like
git add ., instructing the agent to stage only named files or hunks, which reduces the risk of unintended code execution or data leakage. - [COMMAND_EXECUTION]: The skill utilizes shell commands to manage the git lifecycle.
- The skill uses
git stash createandgit update-refto create recoverable snapshots before modifying the repository state. - It employs
git commitandgit pushto automate the PR creation process, governed by explicit instructions to never perform mutations without prior user approval.
Audit Metadata