gh-stack
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the
github/gh-stackextension from GitHub's official extension repository. This is a standard and safe operation for setting up the required tooling. - [COMMAND_EXECUTION]: The skill makes extensive use of
gitandghCLI commands to manage repository state and pull requests. It includes comprehensive instructions for the agent to use non-interactive flags (such as--json,--auto, and--yes) to ensure the agent does not hang on terminal user interfaces (TUIs) or interactive prompts. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from the repository environment, which represents a standard vulnerability surface for developer tools.
- Ingestion points: Data enters the agent's context through
gh stack view --jsonoutput,git loghistory, branch names, and pull request metadata. - Boundary markers: The instructions mandate the use of
--jsonflags for structured parsing, which helps differentiate data from instructions, and provides specific exit codes for unambiguous state checking. - Capability inventory: The skill has the capability to write to the file system (via
git commit), create branches, and perform network operations (viagh stack pushandgh stack submit). - Sanitization: Risk is mitigated by the requirement to parse structured JSON output rather than unstructured terminal text, alongside explicit error code handling.
Audit Metadata