run-retro

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

• Third-party content exposure detected (medium risk: 0.30). The run-retro skill ingests free-form text from “the workstream’s available conversation, repository, issue, pull-request, and CI evidence” (via its workflow steps that build an evidence ledger and invoke grilling on that evidence), which can include outsider-authored content without requiring a specific prior selection by the agent.