Documentation

Everything you need,
explained clearly.

Learn how to manage, version, and deliver structured prompts via API. From sign-up to production in minutes.

Quick Start

Up and running in 4 steps.

From sign-up to your first API call in minutes.

01

Create a Prompt

Sign up, create a project, and add your first prompt. Projects organize prompts by team or application.

02

Add Blocks

Structure your prompt with typed blocks: role, context, instructions, guardrails, output_format, and custom. Drag to reorder.

03

Publish a Version

Lock in your prompt version for production use. Versions are immutable snapshots you can roll back to at any time.

04

Fetch via API

Get your compiled prompt with a single API call. Variables are resolved at fetch time with the values you pass.

bash
$ curl "https://api.promptot.com/api/v1/prompts/my-prompt/compiled" \
-H "Authorization: Bearer pk_live_abc123..."
Core Concepts

How PromptOT works.

Understand the building blocks of structured prompt management.

Blocks

Typed sections (role, context, instructions, guardrails, output_format, custom) that compose into a prompt. Each block has a title, content, type, and sort order.

Compilation

Blocks are filtered (enabled only), sorted by sort_order, and joined as ## Title sections into a single prompt string ready for your LLM.

Variables

Use {{key}} syntax in block content. Variables are resolved at fetch time with values you pass in the API call.

Versions

Publish immutable versions of your prompt. Production API keys always return the published version; development keys return the latest draft.

API Keys

Generate keys scoped to environments (production/development). Keys are hashed with SHA-256 — we never store the raw key.

Webhooks

Get notified when prompts are published, updated, or deleted. Payloads are signed with HMAC-SHA256 for verification.

Questions

Frequently asked questions.

PromptOT is a prompt management platform that lets AI teams compose prompts from typed blocks (role, context, instructions, guardrails, output format, custom), version them safely, and deliver them to applications via a single API call. It supports team collaboration, environment-scoped keys, webhook notifications, and 23 MCP tools for managing prompts from inside AI assistants like Claude Desktop, Cursor, and ChatGPT.
Sign up for free, create a project, add your first prompt with typed blocks, publish a version, and fetch it via the REST API with a single curl command. The entire process takes under five minutes. No credit card required.
PromptOT supports 6 typed block types: role (defines the LLM persona), context (background information), instructions (task directives), guardrails (safety constraints), output_format (response structure), and custom (any additional section). Blocks can be reordered via drag-and-drop and toggled on or off.
Generate an API key scoped to production or development. Make a GET request to /api/v1/prompts/{slug}/compiled with your key in the Authorization header. Production keys return the published version; development keys return the latest draft. Variables are resolved at fetch time with values you pass as query parameters.
Yes. The free tier includes 3 projects, 5 prompts per project, 1,000 API calls per month, 50 AI credits, and up to 5 team members. No credit card required. Pro is $29/month for unlimited projects, prompts, and team members. Enterprise pricing is custom.
Resources

Keep exploring.

API Reference

Complete reference for all public API endpoints, authentication, and error handling.

Changelog

Latest updates, new features, and improvements to the platform.

Status

Real-time platform status and uptime information.

Ready to ship?

Start managing your LLM prompts with structure, versioning, and API delivery. No credit card required.

Get started freeAPI reference