fmt-csv-to-json

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). The skill instructs users to curl and run a JavaScript executable hosted at an unvetted domain (https://fittings.sh/skill/call.mjs), which is a common vector for distributing malicious scripts and is therefore high risk.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md/ call.mjs expose a free-text ingestion path via the user-supplied POST body field csv to https://fittings.sh/v1/fmt/csv-to-json, which the service will parse into structured output (i.e., outsider-authored text is directly consumed as input data by the runtime workflow).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs fetching and then running remote code via "curl -sO https://fittings.sh/skill/call.mjs", which downloads executable JS that would be executed (node call.mjs).

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

• Direct money access detected (high risk: 1.00). The skill endpoint implements an on-chain/off-chain payment flow: it returns a 402 with a signed quote that the caller must sign and resend via a PAYMENT-SIGNATURE header, references a private key (FITTINGS_PRIVATE_KEY), recommends a wallet with USDC on Base, and points to @x402/evm/@x402/core for signing. Those are explicit crypto signing/payment instructions that allow an agent holding a key to authorize real spending, so this grants direct financial execution capability.