pr-finish
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several shell-based commands for repository management, including
git merge,git branch -D, andgh pr merge. It also specifies execution of a local maintenance script,t3-worktree-gc.mjs, to purge build artifacts and manage worktrees. - [INDIRECT_PROMPT_INJECTION]: The skill is instructed to process data from external, potentially untrusted repository files and diagnostic outputs.
- Ingestion points: The skill reads repository-specific documentation such as
AGENTS.md, contribution guidelines, and quality gate documents (as described inSKILL.md). It also processes build, lint, and test error logs. - Boundary markers: Absent. There are no explicit delimiters or instructions to treat the content of external documentation as untrusted data.
- Capability inventory: The skill possesses the ability to delete local and remote branches, merge pull requests via GitHub CLI, and execute local scripts.
- Sanitization: Absent. The skill does not describe any sanitization or validation of the instructions retrieved from the repository's local documentation.
Audit Metadata