nhcx-full
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The specification defines an integration that ingests data from external sources, which creates a vulnerability surface for indirect prompt injection.
- Ingestion points: The integration processes inbound messages from the NHCX exchange (e.g., clinical notes, adjudicator remarks) in "gateway/G8-receive.md" and receives profile data from the ABDM registry in "gateway/G10-beneficiary-registry.md".
- Boundary markers: The pseudocode and FHIR builder logic do not specify the use of delimiters or instructions to treat external data as untrusted content when it is interpolated into the agent's context.
- Capability inventory: The skill provides the agent with capabilities to send data over the network via "gateway/G7-send.md" and perform database writes as described in the "database/" directory.
- Sanitization: There is no explicit requirement or implementation for sanitizing or filtering free-text fields (such as the "remarks" in "database/D25-case-line-item.md") that could contain malicious instructions.
Audit Metadata