Skip to main content
An assistant needs two things to configure your AI agents: the docs, so it knows the API, and an API key, so it can call it.

Give it the docs

The Telzino docs are published in formats AI tools can read:
The docs MCP server only searches and reads the documentation. It can’t read or change your agents. Telzino doesn’t have an MCP server for configuring agents: the assistant changes agents by calling the Telzino API with your key.
Don’t confuse this with MCP integrations, which connect your AI agents to other services during calls.

Give it an API key

  1. A partner creates a key for the assistant under Account Settings → API Keys and scopes it to one reseller or organization with Access. Only partners can create keys. If you’re a reseller or an organization, ask your partner for one. See Overview → Safety rules.
  2. Store the client ID and secret where the tool can read them without showing them in the chat, for example as environment variables:
  3. Delete the key on the API Keys page when you no longer need it.

Setting up common tools

How you connect depends on the tool. In general:
  • Coding assistants that run commands (Cursor, Claude Code and similar): add the docs MCP server in the tool’s MCP settings, and set the two environment variables in the terminal or project the assistant uses. It can then call the API with curl or a short script.
  • Chat assistants (ChatGPT, Claude): add https://docs.telzino.com/mcp as a connector or custom MCP server if your plan supports it, or paste the runbook address into the chat. A chat assistant that can’t make HTTP requests can still read your agent’s settings if you paste them, and write the change for you to apply in the Telzino Dashboard.
Don’t paste a client secret into a chat assistant that keeps conversation history or shares chats with others. Use a tool that reads secrets from the environment, or give the assistant a short-lived key and delete it afterwards.

Starter prompt

Copy this into your assistant and fill in the parts in brackets:
Next, try one of the example prompts.