openapi-directory-first

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

• Third-party content exposure detected (high risk: 0.85). The required workflow fetches public OpenAPI specs from the outsider-run APIs-guru/openapi-directory GitHub repository (via raw.githubusercontent.com URLs), so the LLM ingests outsider-authored free text (YAML/JSON spec content) at runtime.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs fetching OpenAPI specs at runtime (e.g., https://raw.githubusercontent.com/APIs-guru/openapi-directory/main/APIs/{provider.com}/{version}/openapi.yaml and via https://api.apis.guru/v2/list.json or the GitHub contents API) and uses those fetched specs to directly determine prompts/instructions for API calls, so the external content controls agent behavior.