For the complete documentation index, see llms.txt. This page is also available as Markdown.

MCP-Assisted Development

Use GitBook MCP to find workflows, understand endpoints, and speed up Atlas API integration.

💬 Need help? If you're stuck, ask Eva in the Help Center for instant diagnostics.

Ask Eva

Use GitBook MCP to speed up Atlas API development.

It helps you find the right pages, understand the flow, and narrow down the next step faster.

If you want a packaged assistant setup, use Atlas AI Assistant Skill.

When to use MCP

Use MCP when you need to:

  • find the correct API flow for a task

  • understand which identifier is needed next

  • compare similar APIs such as search.do and getOffer.do

  • locate the right troubleshooting page for an error

  • move faster in sandbox without guessing

What MCP is best at

MCP works best as a docs assistant during development.

Use it to:

  • map the booking flow from Search to Pay

  • identify which page explains a field or status

  • find related guides for webhooks, refunds, or utility APIs

  • narrow down which page to open before coding

Start a sandbox build

Ask MCP to outline the minimum sandbox flow.

Then validate against these pages:

Move through the booking flow

Use MCP to confirm the next step after each successful call.

Typical sequence:

  • Search → keep routingIdentifier

  • Verify → keep sessionId

  • Create Order → keep orderNo

  • Payment & Ticketing → poll and confirm final status

Troubleshoot faster

Use MCP to route errors to the right troubleshooting section.

Good starting points:

Prompt patterns that work well

Use short, task-based prompts.

Examples:

  • Show the minimum Atlas sandbox booking flow.

  • I have routingIdentifier. What should I call next?

  • Compare Verify and Get Offer for new integrations.

  • Which pages should I read before implementing ticketing polling?

  • Where should I look for payment decline troubleshooting?

  • Which webhook pages matter after ticketing completes?

Ground rules

Use MCP as a navigation and implementation aid.

Keep these rules:

  • treat API Reference as the source of truth for fields and schemas

  • keep production secrets out of prompts

  • validate request details against your actual environment

  • complete UAT Validation before go-live

Configure your MCP client with this endpoint:

https://resources.atriptech.com/~gitbook/mcp

This endpoint is not a normal web page.

Do not open it directly in a browser.

Suggested workflow

  1. Use MCP to find the right flow.

  2. Open the matching overview or guide page.

  3. Confirm fields and schemas in API Reference.

  4. Build and validate in Sandbox Development.

  5. Use troubleshooting pages when a call or state transition fails.

  6. Move to UAT Validation when the flow is stable.

Last updated

Was this helpful?