fullstack-delivery-loop

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow ingests outsider-authored free text primarily from the OpenSpec change artifacts it “sync[s] the vault” and reads (e.g., proposal.md/specs/ and the OpenAPI contract under <root>/contracts/), which are not authored by the operating user and can contain arbitrary prose that the orchestrator then places into the LLM context as part of planning and dispatch.