fittings-security

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

• Third-party content exposure detected (medium risk: 0.30). Runtime workflow reads outsider-provided free text from user prompts to build and execute targeted queries against endpoints like cve.lookup/cve.search and screening calls (e.g., csl.name-screen) on https://fittings.sh, where the LLM ingests user-authored identifiers/names without first selecting a specific pre-authorized item.

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

• Direct money access detected (high risk: 1.00). The skill explicitly requires a spending wallet private key (FITTINGS_PRIVATE_KEY) and specifies payment via USDC on Base, using an off-chain signature that authorizes a USDC transfer (the facilitator submits the transaction). Those statements (exporting a private key, instructing it be used to pay, and describing the transfer mechanism) indicate the skill can sign/authorize crypto payments and thus perform direct financial execution.