pr-loop
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Interacts with the
ghcommand-line utility to retrieve pull request details and status from GitHub. - [COMMAND_EXECUTION]: Uses shell redirection to create and update JSON state files in the
/tmpdirectory, incorporating the pull request identifier into the file path. - [PROMPT_INJECTION]: Processes content from external GitHub Pull Request comments, creating a surface for potential indirect prompt injection attacks.
- Ingestion points: Fetches and acts upon pull request comments using the
respond-to-commentsskill. - Boundary markers: Does not specify the use of delimiters or 'ignore' instructions when handling external comment text.
- Capability inventory: Able to perform shell commands, write to the local filesystem, and modify recurring session-scoped jobs.
- Sanitization: The instructions do not define sanitization or filtering procedures for the ingested comment data.
Audit Metadata