Ship AI prompts like
you ship code.

Manage prompts for your AI agents and apps in production. Version control, rollback, and team collaboration without redeploying.

Get Started FreeSee how it works
Products

Everything you need to manage prompts

A full editor with versioning, variables, test cases, and team collaboration.

01

Prompt Editor

Multi-message editor with system, user, and assistant roles. Variables with {{syntax}} get auto-detected, highlighted, and tracked with metadata.

02

Publish Workflow

Edit prompts as drafts. When you're ready, hit publish. Your app picks up the new version instantly via API. Full version history with one-click rollback.

03

Test Cases & Preview

Save named test cases with different variable values. Preview AI responses right in the editor before going live.

04

Evaluate & Compare

Run your test cases against draft changes. See scores side-by-side across versions. Catch regressions before they reach production.

GitHub Integration

Already have prompts
in your code?

Connect your repo. We scan for hardcoded prompts, show you exactly what to change, and create a PR replacing them with SDK calls. Your production code stays untouched until you merge.

Scans JS, TS, and Python files automatically
Shows before/after diff for every prompt
One-click PR creation, no manual code changes
Detects when source files change after import
SDK & API

Prompt to production in 5 lines

Fetch your prompt. Compile variables. Send to any LLM.That's your entire integration.

One SDK. Any LLM.

Install the SDK, fetch your prompt by slug, compile variables, and pass the messages to OpenAI, Anthropic, Google, or any provider. Your prompts live in Montage. Your code just fetches them.

TypeScript & Python SDKs
Built-in caching
Variable compilation
import { Montage } from '@montage-sh/sdk'

const montage = new Montage({ apiKey: 'mt_live_...' })

const prompt = await montage.get('research-agent')
const compiled = prompt.compile({ task: 'Analyze Q1 metrics' })

// Use with any LLM
openai.chat.completions.create({
  messages: compiled.messages
})
AI-Native Workflow

Your AI writes the integration.
Your team owns the prompts.

The bottleneck

Right now, changing a prompt means changing code, opening a PR, and deploying. Your PM has a better idea for the support bot's tone, but they need an engineer to ship it.

The fix

Tell Claude Code, Cursor, or Codex to set up Montage. It moves your prompts out of code and into a shared workspace. Wired, versioned, and ready. Now anyone can edit, test, and ship prompt changes without waiting on engineering.

Engineers stay in control with approvals, versioning, and rollback.

Your AI tool

> Move our prompts into Montage
  so the team can manage them.

Claude: Setting up Montage...
✓ Installed SDK
✓ Found 4 prompts in codebase
✓ Synced to workspace

$ montage publish --all
✓ research-agent-planner (v1)
✓ code-review-system (v1)
✓ support-agent-tools (v1)
✓ data-extraction (v1)

Stop shipping prompt changes through pull requests

Let PMs and operators update AI behavior instantly. No IDE, no deploy, no waiting on engineering.

Catch bad outputs before they hit production

Test prompt variations with saved test cases. Compare outputs side-by-side. Know which version is better before it goes live.

See every prompt your product uses

No more digging through codebases. Every prompt, variable, and version lives in a workspace your whole team can actually use.

Roll back mistakes in seconds, not deploy cycles

Every publish is a version snapshot. Something breaks? One click to revert. Full diff view shows exactly what changed.

Built for teams

Your whole team,
one prompt platform.

Engineers, PMs, and content writers all edit prompts in one place. Approval workflows keep production safe.

Approval workflows

Require reviews before prompts go live. Configurable per-project with approve, reject, and request-changes flows.

Montage is
built for

Engineering Teams

Update prompts without redeploying. Version control and rollback for every change. GitHub integration scans your repo and replaces hardcoded prompts with SDK calls.

Product & Content Teams

Edit AI behavior without writing code. Approval workflows ensure changes are reviewed before going live. Comments and @mentions keep everyone aligned.

AI Startups & SaaS

Ship faster with remote prompt config. Go from hardcoded strings to managed prompts in minutes. Evaluate changes with test cases before they reach users.

AI Agent Teams

Manage prompts for autonomous agents from one dashboard. Update planning, tool-use, and persona prompts across your agent fleet. Roll back when an agent goes off-script.

Agencies & Consultancies

Manage prompts across client projects with organizations and teams. Role-based access at every level. Analytics show exactly how each prompt performs.

00

FAQ: Prompt Management

Pricing

Simple, transparent pricing

Start free. Upgrade when you need more.

Free

$0forever

For side projects and experiments.

  • 3 projects
  • 10 prompts per project
  • 1,000 API calls / month
  • Version history
  • Community support
Get Started

Pro

Popular
$10/ month

For developers shipping AI to production.

  • Unlimited projects
  • Unlimited prompts
  • 100,000 API calls / month
  • Analytics & observability
  • Test case auto-generation
  • Priority support
Start Pro

Organization

$25/ month

For teams that need collaboration and control.

  • Everything in Pro
  • 1,000,000 API calls / month
  • Teams & role-based access
  • Approval workflows
  • Org-level analytics
Start Org

Ready to stop hardcoding prompts?

Import your existing prompts in seconds. No credit card required.

Free tier includes 3 projects, 10 prompts each, 1,000 API calls/month.

Get Started Free