eas-workflows

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

• Third-party content exposure detected (high risk: 0.75). The required workflow fetches public web content at runtime (Expo’s JSON schema and MDX syntax/docs via node <skill-dir>/scripts/fetch.js <url> and scripts/validate.js calling fetchCached), and that fetched free-form text/JSON is then parsed and used to generate/validate responses in 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). The skill fetches the runtime schema and docs (notably https://api.expo.dev/v2/workflows/schema and the raw.githubusercontent.com MDX URLs) via scripts/fetch.js and scripts/validate.js, and that fetched content is used as the authoritative source to validate workflows and to derive answers, so remote content directly controls agent outputs.