ship
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of shell commands through
gitand the GitHub CLI (gh) to perform repository operations such as fetching tags, diffing branches, committing changes, and managing pull requests. - [DYNAMIC_EXECUTION]: The skill executes arbitrary commands defined within the repository's own configuration files (specifically in
docs/agents/). This includes 'Fast checks', 'Bump command', 'Next-version command', and 'Changelog check'. While these are intended for project-specific automation, they allow for execution patterns determined at runtime by the workspace environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data that could contain malicious instructions designed to influence the agent's behavior during the shipping process.
- Ingestion points: Untrusted data is ingested from issue bodies and comments via
gh issue view, from pull request diffs viagh pr diff, and from project-specific settings files indocs/agents/*.md. - Boundary markers: The instructions do not define explicit boundary markers or 'ignore' instructions for data retrieved from issues or diffs.
- Capability inventory: The skill possesses significant capabilities, including the ability to write to the local filesystem (changelogs and docs), execute shell commands defined in configuration, and push code to remote repositories (
git push). - Sanitization: While the skill includes procedures to prevent the accidental commitment of secrets (e.g., checking for private-key headers), it lacks robust sanitization or validation for the natural language content processed from issue trackers and diffs.
Audit Metadata