The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves updates, manifest files, and documentation zips from the vendor's repository at github.com/nha-in/nhcx-package. These are documented as vendor-owned resources used for maintaining compliance with health exchange standards.
[REMOTE_CODE_EXECUTION]: Instructions in SKILL.md direct the agent to periodically download updated versions of the skill files from GitHub and replace its own local folder. The process includes a verification step using SHA-256 checksums to ensure file integrity.
[INDIRECT_PROMPT_INJECTION]: The skill ingests specifications and FHIR examples from a remote 'knowledge source' package extracted at runtime.
Ingestion points: Documentation and examples loaded from nhcx-plan/knowledge/ after downloading from GitHub.
Boundary markers: Absent; the agent is instructed to follow instructions in these specifications without explicit separation from original instructions.
Capability inventory: File system modification (nhcx-plan/), network operations via the NHCX gateway, and generation of integration scripts.
Sanitization: Ingested documentation is checked for integrity but not filtered for potential instruction injection in natural language text.
[DYNAMIC_EXECUTION]: The skill instructs the agent to generate and execute a reporting script (report.py or similar) to visualize the integration progress based on the contents of the generated plan JSON files.

nhcx-payer-claim