Repolode
Pomerado

Pomerado/pomerado

Builds MCP integrations for websites from a plain-language task, so AI agents can call sites that have no API.

Stars
1,573
Forks
135
Open issues
29
Last commit
2 hours ago

What is it?

Pomerado turns a website into an MCP server that an AI agent can call. You describe the task in plain language, such as reading a page heading or submitting a form, and Pomerado drives a real Chromium browser to explore the site, complete an example, and generalize it into deterministic source code. The generated integration is saved as a folder you own and runs without a model call.

The approach aims to be more reliable than having an agent operate a browser through computer use on every step. Pomerado also handles sites behind a login and two-factor authentication. Integrations are scoped as read-only or write, and a guardian component reviews each browser action and the final source during minting.

The core is MIT-licensed and runs locally with Node 24 on macOS or Linux, using an OpenAI API key for minting. It works with MCP clients such as Claude Code, Codex, Cursor, VS Code, Claude Desktop and Gemini CLI.

Key features

  • 01Builds integrations from a natural language task description, with no HAR file or recording needed.
  • 02Runs generated integrations deterministically, without a model call per request.
  • 03Supports sites behind login and two-factor authentication.
  • 04Separates read and write authority so integrations can be limited to viewing only.
  • 05Saves each integration as editable source code in a local folder.
  • 06Ships a standard MCP stdio server with mint, get_job, provide_input and cancel_job tools.

Use cases

  • Giving an AI agent read access to a website that offers no API.
  • Automating a repeated form submission or booking on a logged-in site, with write authority.
  • Exposing an internal web tool to MCP clients as a callable tool.
  • Prototyping site integrations before deciding whether to build an official API client.

Quick start

  1. 1

    Install Pomerado and the Chromium browser it drives

    npx -y -p pomerado pomerado-mcp --help
  2. 2

    Install Chromium for Playwright

    npx -y -p pomerado playwright install chromium
  3. 3

    Add the Pomerado MCP server entry to your MCP client config (see README for per-client commands)

Things to know

  • Minting an integration requires an OpenAI API key.
  • Requires macOS or Linux and Node 24.21 or a later Node 24 release.
  • Pomerado Cloud features such as hosting, saved logins and bot protection are not part of the open-source core.

In collections

Topics

FAQ

What is pomerado?
Builds MCP integrations for websites from a plain-language task, so AI agents can call sites that have no API. Pomerado turns a website into an MCP server that an AI agent can call. You describe the task in plain language, such as reading a page heading or submitting a form, and Pomerado drives a real Chromium browser to explore the site, complete an example, and generalize it into deterministic source code. The generated integration is saved as a folder you own and runs without a model call.
Is pomerado free to use? What license does it have?
pomerado is open source under the MIT license. Check the license terms before using it in commercial products.
Is pomerado still maintained?
Actively maintained. The last commit to pomerado was on Nov 7, 2026.
What can I use pomerado for?
Giving an AI agent read access to a website that offers no API. Automating a repeated form submission or booking on a logged-in site, with write authority. Exposing an internal web tool to MCP clients as a callable tool. Prototyping site integrations before deciding whether to build an official API client.
How do I install pomerado?
Install Pomerado and the Chromium browser it drives: npx -y -p pomerado pomerado-mcp --help Install Chromium for Playwright: npx -y -p pomerado playwright install chromium Add the Pomerado MCP server entry to your MCP client config (see README for per-client commands)
Where can I get pomerado?
The source code and installation instructions are on GitHub: https://github.com/Pomerado/pomerado

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects