fittings-text-profanity-filter

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). These documents instruct you to curl and execute an unverified JavaScript file (call.mjs) served directly from fittings.sh — downloading and running remote scripts from an untrusted service is a high-risk vector for malware or credential theft.

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow is a direct, user-supplied POST to https://fittings.sh/v1/text/profanity-filter where the provided text is read and processed for profanity matches, so outsider-authored free text is ingested.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's instructions tell the runtime to curl and save remote JavaScript (https://fittings.sh/skill/call.mjs) and then execute it with node, which fetches and runs external code at runtime.

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

• Direct money access detected (high risk: 1.00). The skill includes an explicit crypto payment flow: the service returns a signed quote, the caller "signs an authorization", and the docs show setting FITTINGS_PRIVATE_KEY (export 0x...) and running call.mjs to pay with USDC on Base. Those instructions and environment-variable private-key usage are explicit wallet/signing steps that enable an agent to perform on-chain/crypto payments, so this grants direct financial execution capability.