AI Onboarding

If you build your Partner API integration with an AI coding agent such as Claude Code, Cursor, Codex or GitHub Copilot, point it at these docs. Every page is available as plain markdown, so the agent reads the same API reference you do, always up to date.

Prerequisite: Create an API KeyCopy Anchor Link

A member of your partner account creates the API key and assigns the capabilities it needs — see API Keys. Give the key to your agent through an environment variable, never by pasting it into a chat.

Docs for AgentsCopy Anchor Link

1. Markdown pages. Add .md to any page URL, or request the page with an Accept: text/markdown header.

   Copy Text

   https://portal.peplink.com/docs/api-reference/getting-started.md

2. llms.txt. An index of every page, with a one-line summary and a link to its markdown version.

   Copy Text

   https://portal.peplink.com/docs/llms.txt

3. llms-full.txt. All the docs in a single file.

   Copy Text

   https://portal.peplink.com/docs/llms-full.txt

You can also use Copy Markdown and Open in at the top of any page to hand that page to ChatGPT, Claude or another assistant.

Prompt for Your AgentCopy Anchor Link

Start your agent with this prompt. It sends the agent to this page, whose markdown version includes the rules below.

Read https://portal.peplink.com/docs/ai-onboarding.md and follow its rules for agents. Then build: (describe your integration here)

OpenAPI Spec and SDKsCopy Anchor Link

We don't publish an OpenAPI (Swagger) specification or client SDKs. The API Reference is the contract. Each endpoint's page lists its method and path, parameters, required capabilities, every response field with its type and enum values, and a full example request and response. That is all an agent needs to write a typed client for the endpoints your integration uses.

Rules for AgentsCopy Anchor Link

1. Read Getting started first. It covers authentication, the request URL, pagination, array and nullable query parameters, and date formats. These apply to every endpoint.
2. Use only what is documented. Fetch an endpoint's markdown page before calling it and use its exact path, parameters and field names. Don't infer endpoints or fields from other APIs. If something the integration needs isn't documented, stop and tell the developer. Undocumented behavior is not supported.
3. Keep the API key out of code. Send it as Authorization: Bearer <api-key>, read from an environment variable. Never hard-code, log or commit it.
4. Check Required Capabilities. If an endpoint's page lists them, the API key must have them assigned. When a call is rejected, check this first.
5. Page through search results. Search endpoints take pageNum (from 1) and pageSize (up to 100) and return totalCount with the current page's items. Keep requesting the next page until you have totalCount items.
6. Ask before writing data. The API key acts on the partner's live Peplink account. Retrieve, search and list calls are safe to explore with. Ask the developer before any call that creates or changes data, such as orders, sales registrations, SIM suspend and resume, tags, webhooks, and Starlink requests.
7. Don't blindly retry writes. If a request that creates data times out or returns a 5xx, check whether it took effect (for example, retrieve the order) before sending it again. A 4xx means the request needs fixing; don't resend it unchanged.
8. Follow a recipe when one exists. Recipes walk through multi-step flows such as placing an order or activating a Starlink service line.

Keeping Up to DateCopy Anchor Link

API changes are announced in the Changelog. Have your agent check it before updating an existing integration.