fittings-doc-url-to-markdown

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). The documentation instructs downloading and executing a remote JavaScript module (call.mjs) from fittings.sh, which is a high-risk pattern because running externally fetched scripts can execute arbitrary/malicious code.

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

• Third-party content exposure detected (medium risk: 0.30). The skill doc.url-to-markdown at https://fittings.sh/v1/doc/url-to-markdown ingests caller-provided free text via the url field and then performs a live fetch of that public web page, returning its readable content as Markdown.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill calls the external endpoint https://fittings.sh/v1/doc/url-to-markdown at runtime to fetch caller-supplied web pages (Source: "live fetch of the caller-supplied URL"), which will return arbitrary text/Markdown that can directly influence agent prompts and behavior.

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

• Direct money access detected (high risk: 1.00). The skill requires on-chain-style payment authorization: it instructs users/agents to set a private key (FITTINGS_PRIVATE_KEY), sign an authorization, and uses a wallet holding USDC on Base; the client libraries mention EVM tooling (@x402/evm). These are explicit crypto wallet/signing/payment integration details (signing to authorize settlement), which constitute Direct Financial Execution capability.