doc-url-to-markdown

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

• Third-party content exposure detected (medium risk: 0.30). The runtime path /v1/doc/url-to-markdown accepts a caller-supplied url and performs a live fetch of that external page, then ingests the fetched page text (converted to Markdown) for output—so outsider-authored content can be supplied to the workflow via the request body.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill calls the remote service endpoint GET https://fittings.sh/v1/doc/url-to-markdown which at runtime fetches an arbitrary caller-supplied public URL and returns its cleaned Markdown (i.e., external web content is injected into the agent's context and can directly control prompts/responses).

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill describes an explicit crypto-based payment flow: the endpoint returns a 402 with a signed quote, the client must "sign an authorization" using a private key (FITTINGS_PRIVATE_KEY), and it references USDC on Base and EVM signing libraries (@x402/evm). These are concrete wallet/signing instructions that enable on‑chain/crypto payment authorization, so the skill includes direct crypto/financial execution capability.