acp-capability-negotiation

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md "Before writing code" steps explicitly require fetching live docs from public sites (e.g., web-searching GitHub for the capability_negotiation RFC and fetching https://developers.openai.com/commerce/specs/checkout/), which means the agent will ingest and act on untrusted, user-provided third-party content that can materially change implementation and behavior.