checking-merge-readiness
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a processing engine for untrusted data fetched from external pull requests. This content is interpolated into the agent's reasoning process to determine merge readiness.
- Ingestion points: External data is gathered via
gh pr view(identity, body, comments),gh pr diff(code changes), andgh issue view(linked issues). - Boundary markers: The skill contains explicit instructions to treat all forge-derived text as untrusted and specifically includes a driver for 'Assessment steering' in the risk rubric to detect and grade attempts to influence the agent's verdict.
- Capability inventory: The skill possesses the ability to read local files, execute shell commands (GitHub CLI), and perform write operations to the forge (merging pull requests).
- Sanitization: The implementation uses GraphQL variables for API interactions and processes data using
jqto ensure that untrusted text does not flow into command arguments. - [COMMAND_EXECUTION]: The skill relies on executing system commands and a bundled bash script to perform its duties.
- Evidence:
SKILL.mdandreferences/merge-execution.mddetail the execution ofgh,jq,shasum, and the local scriptscripts/fetch-pr-history.shto fetch history, check status, and perform the final merge.
Audit Metadata