finishing-a-development-branch
Pass
Audited by Gen Agent Trust Hub on Apr 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes shell commands for git operations (checkout, pull, merge, branch) and project-specific test runners (npm, cargo, pytest, go). These operations are essential for the skill's workflow management purpose.\n- [EXTERNAL_DOWNLOADS]: Communicates with remote repositories through git pull/push and creates pull requests using the GitHub CLI (gh). These operations are restricted to repositories and services already configured by the user.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted metadata from the project environment.\n
- Ingestion points: Git branch names, commit messages, and output from local test suites.\n
- Boundary markers: The skill employs a quoted heredoc ('EOF') when constructing the Pull Request body, which prevents local shell expansion of variables but does not prevent the agent from reading potentially malicious instructions in the metadata.\n
- Capability inventory: Shell command execution via git and the GitHub CLI.\n
- Sanitization: No explicit sanitization or filtering of git metadata or test output is performed before it is presented to or used by the agent.
Audit Metadata