prow-trigger-nightly

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The skill fetches outsider-authored public web content at runtime from https://prow.ci.openshift.org/configured-jobs/<repo> (HTML containing embedded JSON job names), which is then parsed and included in the agent’s displayed context during --list/job selection.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The script performs runtime fetches that directly control user prompts: it queries Prow job pages (e.g. https://prow.ci.openshift.org/configured-jobs/redhat-developer/rhdh) to build the selectable job list and queries the Quay API (e.g. https://quay.io/api/v1/repository/rhdh%2Frhdh-hub-rhel9/tag/) to populate tag options, so external content directly controls the agent's prompts.