Admin API Overview

What Can You Do?

Manage Endpoints

Create, update, and delete endpoints. Define input and output schemas programmatically.

Manage Prompts

Create prompts, update templates, promote to Live, and manage the full prompt lifecycle.

Execute & Test

Execute endpoints with inputs and get LLM outputs — ideal for automated testing.

Monitor & Analyze

Query execution stats, browse logs, and compare prompt performance.

Interface	Best For	Protocol
REST API	Scripts, CI/CD, backend integrations	Standard HTTP/JSON
MCP Server	AI coding tools (VS Code, Claude, Cursor)	Model Context Protocol

Interface

Best For

Protocol

REST API

Scripts, CI/CD, backend integrations

Standard HTTP/JSON

MCP Server

AI coding tools (VS Code, Claude, Cursor)

Model Context Protocol

Quick Start

Create an Admin API Key

Go to API Keys in your dashboard and create a key with the Admin type. Learn more →

Choose Your Interface

Use the REST API for scripts and integrations, or the MCP Server to connect your AI coding tool.

Start Building

Create an endpoint, add fields, write a prompt, promote it to Live, and execute it — all without touching the dashboard.

Example: Create and Deploy an Endpoint

Here’s the typical workflow using the Admin API:

Create an endpoint         POST /admin/v1/endpoints
Add input fields           POST /admin/v1/endpoints/{id}/input-fields
Add output fields          POST /admin/v1/endpoints/{id}/output-fields
Create a prompt            POST /admin/v1/prompts
Promote to Live            POST /admin/v1/prompts/{id}/promote
Set as default             POST /admin/v1/prompts/{id}/set-default
Execute                    POST /api/v1/{endpoint-path}

The same workflow works with MCP tools — AI coding assistants can execute these steps conversationally.

Resources

The Admin API manages these resources:

Resource	Description
Endpoints	API paths with input/output schemas, cache and rate-limit policies
Prompts	Liquid templates with LLM configuration, versioning, and lifecycle
Prompt Versions	Immutable snapshots created automatically when templates change
Snippets	Reusable template fragments included via `{% include 'slug' %}`
Execution Logs	Read-only history with inputs, outputs, latency, tokens, and cost
Stats	Aggregated execution metrics per endpoint, prompt, or tenant
Models	Available LLM models with capabilities and pricing

​What Can You Do?

Manage Endpoints

Manage Prompts

Execute & Test

Monitor & Analyze

​Two Ways to Access

​Quick Start

​Example: Create and Deploy an Endpoint

​Resources

What Can You Do?

Two Ways to Access

Quick Start

Example: Create and Deploy an Endpoint

Resources