api-tester

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

• Third-party content exposure detected (high risk: 0.75). The workflow’s runtime calls to v1/proxy/orthogonal/run include user-supplied website_url/url fields (e.g., https://api.example.com/docs), which can cause the system to fetch and parse outsider-authored public web content and feed it into the agent/LLM context via the proxy/scrapegraph/linkup paths.

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

• Potentially malicious external URL detected (high risk: 0.90). This skill makes runtime requests to fetch and scrape external documentation (e.g., https://stripe.com/docs/api and https://api.openai.com/v1/models via the Gooseworks proxy) and then injects that fetched content into prompts/instructions to generate tests, so those external URLs directly influence agent behavior.