New MCP integration for AI coding agents

Every pull request
deserves a live environment

Auto-tester gives your AI agents and CI pipelines the power to deploy preview environments instantly. One config. One command. A live URL in your PR, every time.

Get startedView on GitHub
.github/workflows/preview.yml
name: Preview Environment
on:
  pull_request:
    types: [opened, synchronize, reopened, closed]

jobs:
  preview:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: previo/previo@v1
        with:
          api-key: ${{ secrets.AUTO_TESTER_API_KEY }}
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

From commit to live URL in seconds

The entire lifecycle is automatic. You write code. Auto-tester handles everything else.

01

Agent opens a PR

Your AI coding agent pushes a branch and opens a pull request. The GitHub Action triggers automatically.

02

Environment spins up

Auto-tester builds your docker-compose stack, assigns a unique URL, and streams logs in real time.

03

Link appears in the PR

A live preview URL and log viewer link are posted as a comment. Click and test immediately.

04

PR merges, environment vanishes

When the PR closes, the environment is destroyed. Zero cleanup. Zero orphaned infrastructure.

Built for the way you already work

Not another tool to learn. Auto-tester meets you where you are: in your terminal, your CI, and your agent's toolchain.

MCP-native

AI agents don't read dashboards. They call tools. Auto-tester ships a full MCP server with seven tools for the complete environment lifecycle.

mcp-tools
create_environment()
deploy_environment()
get_environment_status()
get_environment_logs()
destroy_environment()

Real-time logs

WebSocket-powered log streaming from build to runtime. Every container, every line, the moment it happens. Share the link and your team sees exactly what the agent sees.

previo logs pr-42
18:42:01 [INFO] [system] Starting deployment...
18:42:03 [INFO] [web] Building image...
18:42:18 [INFO] [web] Container started
18:42:19 [INFO] [db] PostgreSQL ready
18:42:20 [INFO] [system] Environment running
         -> https://preview.dev/env/pr-42

Self-host in one command

No vendor lock-in. Run previo on your own infrastructure with a single docker compose up. Your code, your servers, your control.

bash
$ cp .env.example .env
$ docker compose up -d

  server    ✓ running  :8080
  web       ✓ running  :3000
  postgres  ✓ running  :5432
  caddy     ✓ running  :443

Developer experience is the product

We obsess over every interaction so you can stay in flow.

Typed SDK

Full TypeScript client with autocomplete for every endpoint.

CLI

previo preview from your terminal. Instant feedback loop.

GitHub Action

Five lines of YAML. That's the entire integration.

API-first

Every feature accessible via REST. Build your own workflows.

Live log viewer

WebSocket-powered dashboard. No refresh. No polling.

Zero config

Bring your docker-compose.yml. Auto-tester figures out the rest.

Ship faster.
Review live code.

The future of code review isn't reading diffs. It's clicking a link and experiencing the change.

Start building