keyapi-amazon

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

• Third-party content exposure detected (high risk: 0.75). High: the required workflow fetches public KeyAPI docs at runtime (e.g., https://docs.keyapi.ai/llms.txt and linked endpoint pages) via scripts/search-keyapi-docs.mjs, and the fetched docs text is parsed and then used to construct the LLM-visible context (method/path/params/body/response contract) before live calls.

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

• Potentially malicious external URL detected (high risk: 0.90). The helper script scripts/search-keyapi-docs.mjs fetches https://docs.keyapi.ai/llms.txt at runtime (and may resolve linked docs pages) to extract the current OpenAPI method/path/parameters, which directly controls the agent's prompt-to-action routing and subsequent requests.