setup-engine

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime path includes WebSearch/WebFetch to look up “latest stable [engine] version” and to fetch official migration/breaking-change docs, which can ingest outsider-authored free text from public web pages into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). This skill explicitly uses WebSearch/WebFetch at runtime to fetch official migration/matching-engine documentation (e.g., migration guide URLs discovered via searches such as "[engine] [old-version] to [new-version] migration guide") and then parses and injects that fetched content into its reference docs and agent instructions, so runtime web URLs (the migration guide / docs URLs returned by WebFetch) directly control agent behavior.