git-pr-workflows-git-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes uncommitted code changes, file lists, and PR metadata which are treated as untrusted data that could contain malicious instructions designed to bypass agent constraints.
- Ingestion points: Phase 1 (Code Quality Assessment) and Phase 3 (Change Analysis) involve reading uncommitted changes and file content from the local repository.
- Boundary markers: The instructions do not define clear delimiters or use specialized instructions to tell the subagents to ignore potential natural language instructions embedded within the code being analyzed.
- Capability inventory: The workflow utilizes subagents like
unit-testing::test-automatorandcicd-automation::deployment-engineerwhich are instructed to execute shell commands for running tests and managing Git branches/pushes. - Sanitization: There is no evidence of sanitization, escaping, or validation of the ingested code content before it is processed by the subagents.
- [COMMAND_EXECUTION]: The skill directs the agent to perform operational tasks that involve shell execution, specifically running unit, integration, and end-to-end tests, as well as performing Git operations like branching and pushing to remotes.
Audit Metadata