github-workflow-automation

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text is ingested into the LLM via GitHub event payloads (e.g., issue.body / PR review comment text) in the “Issue Triage Automation” and “@mention Bot” workflows, where issue.title, issue.body, and github.event.comment.body are interpolated into the AI prompt at runtime.

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

• Potentially malicious external URL detected (high risk: 0.70). The workflows call third-party GitHub Actions that are fetched and executed at runtime (e.g., actions/github-script -> https://github.com/actions/github-script, actions/checkout -> https://github.com/actions/checkout, actions/stale -> https://github.com/actions/stale, slackapi/slack-github-action -> https://github.com/slackapi/slack-github-action), so remote repository code will run as part of the skill and thus represents an external runtime dependency that can execute code.