nhcx-full

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

• Third-party content exposure detected (medium risk: 0.30). The workflow consumes outsider-submitted free text via policy search inputs and benefit claim discovery identifiers retrieved from the external beneficiary identification system.

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

• Potentially malicious external URL detected (high risk: 0.70). The instruction mandates downloading and installing a remote skill package from GitHub via a specified raw URL or release archive if a manifest version is newer, which constitutes an operational runtime external dependency from an unofficial/unknown provenance tier.

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

• Direct money access detected (medium risk: 0.30). The skill provides instructions for building provider-side integration for India's NHCX (National Health Claims Exchange) into a hospital management information system (HMIS), including policy search, coverage eligibility, pre-authorisation, claims submission, payment notices, and related workflows. This involves indirect financial operations (submitting claims and billing usage/requests for downstream handling) and direct financial operations (payment notices, settlement release). However, all operations and API calls are standard FHIR-based healthcare exchange protocols managed via participant credentials, encryption certificates, and gateway configurations without establishing or executing direct end-user money transfers, charges, or purchases within the skill's own operational scope. Consequently, this falls under indirect financial operations, resulting in a medium severity rating.