Skip to main content
An LLM — Claude, ChatGPT, Cursor, or your own agent — can speed up your Nash integration considerably: generating request code, explaining endpoints, and debugging responses. This page covers the formats we publish for exactly that.

​
Nash API Documentation for LLMs

Give the LLM the API documentation in a machine-readable format: These specially formatted text files contain the Nash API endpoints, schemas, and implementation details in a format that’s optimized for LLMs to process. They’re generated from the public Nash API specification, so they describe only the endpoints available to you — no internal surfaces.

​
Example Prompt

I'm working on integrating with Nash's delivery API. Here's the relevant documentation:
https://docs.usenash.com/llms-full.txt

I need to implement [SPECIFIC FUNCTIONALITY]. Can you help me understand how to structure this request?

​
Best Practices for Using LLMs with Nash API

  1. Provide Clear Context: Share the relevant Nash API documentation links with the LLM
  2. Be Specific: Clearly describe your use case and requirements
  3. Include Complete Information: When troubleshooting, share error messages and API responses

​
Key Integration Areas Where LLMs Can Help

  • Understanding Nash Concepts: Orders, Jobs, Batch Jobs, Dispatch Strategies
  • API Request/Response Formatting: Generate code samples for API calls
  • Webhook Implementation: Set up secure webhook handlers for delivery updates
  • Error Handling: Understand and resolve API error responses

​
Keep sensitive data out of the prompt

The documentation links above are safe to share — they’re public. Your operational data is not. Treat anything you paste into a third-party LLM as if it could be retained or used for training, and keep these out of the prompt:
  • Credentials — API keys, tokens, Authorization headers, webhook signing secrets.
  • Customer and recipient PII — names, dropoff addresses, phone numbers, email addresses.
  • Identifiers that map back to your operation — organization IDs, Nash-Org-Id, order/job IDs, request IDs.
  • Raw API responses and webhook payloads — these bundle several of the above together. Redact before sharing.
When you need an LLM to debug a real request or response, replace the sensitive values with placeholders (<API_KEY>, <ORG_ID>, 123 Main St) before pasting. The structure is what the model needs; the real values are not.
Always review and test LLM-generated code before deploying it to production — verify endpoints, fields, and error handling against the API reference.
For comprehensive integration guidance, refer to our Plan Your Integration documentation.