promoting-a-release
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads instructions and commands from a project-level file (
docs/agents/promotion.md) and executes them in the local environment. - Ingestion points: The skill ingests command strings for testing, database migrations, building, and smoke checking directly from the
docs/agents/promotion.mdfile within the repository. - Boundary markers: The instructions do not define boundary markers or safety delimiters to isolate the content of the promotion file from the agent's core instructions.
- Capability inventory: The skill possesses the capability to execute arbitrary shell commands via the system shell, as well as high-level version control operations using
gitandghtools. - Sanitization: There is no evidence of validation, escaping, or sanitization of the command strings retrieved from the configuration file prior to execution.
- [COMMAND_EXECUTION]: The skill relies on executing shell commands to facilitate the release process.
- Evidence: The skill explicitly executes
git fetch,git checkout,git merge, andgh pr merge, along with dynamic commands likenext buildor migration scripts provided by the user in the repository configuration.
Audit Metadata