tailor-to-venue

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

• Third-party content exposure detected (high risk: 0.75). SKILL.md’s required runtime workflow fetches and re-verifies the target venue’s live CFP (via parse-cfp/CFP URL), so outsider-authored free text from the public CFP page can be ingested into the agent/LLM context through the CFP-derived verification record.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs fetching the venue's live CFP (the profile's cfp_url from venues/conferences/.yml / profile.get("cfp_url")) at runtime to re-verify requirements, and that fetched content is used to drive the tailoring plan, so the live CFP URL is a runtime external dependency that directly controls agent behavior.