pr-watch-fix
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands through git, gh, and make to perform repository operations, fetch CI status, and update generated files. These actions are aligned with the skill's intended functionality for PR automation.\n- [DATA_EXFILTRATION]: Network communication is performed to interact with GitHub APIs and push commits to remote branches. These operations target a well-known service (GitHub) and are necessary for the skill's operation.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via processed CI logs.\n
- Ingestion points: CI failure logs are retrieved from GitHub Actions using the
gh run viewcommand and stored in/tmp/pr-watch-fail.logfor analysis.\n - Boundary markers: The instructions do not define explicit boundary markers or directives to ignore potentially malicious content within the logs.\n
- Capability inventory: The skill can edit source files, execute build scripts via
make, and push code changes to the remote repository.\n - Sanitization: Content extraction from logs is performed using
grepandsedto identify specific error patterns, but there is no formal validation or sanitization of the raw log data.
Audit Metadata