Oya Browser MCP
A browser for agents and people with an MCP server: the agent works in tabs with your logins while you watch and can take over
- 351 ★ on the repository
- Last commit
Broad access
Why this level
- The agent acts in a browser with your imported logins and can perform actions as you on websites
- Pages the agent reads may contain instructions aimed at it, so keep sign-in, payment and email-sending steps for a person
- The API key opens access to your Oya account and its cloud browsers
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/oya-browserDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent acts in a browser with your imported logins and can perform actions as you on websites
- Pages the agent reads may contain instructions aimed at it, so keep sign-in, payment and email-sending steps for a person
- The API key opens access to your Oya account and its cloud browsers
This is third-party code. Review the repository files before installing.
What it does
Oya Browser combines a desktop app, cloud browsers and a control server. An agent can work in your already open browser with imported logins, start managed cloud browsers, or connect through familiar tools and APIs. The agent's actions run in separate tabs, and an AI Mode and Human Mode switch lets you take over. There are three ways to connect it to assistants. The desktop app pairs with Claude Code, Codex, Cursor and VS Code. The remote MCP server runs in Oya's cloud and takes an API key in the Authorization header. For code there is a Node.js SDK and a CLI, which can also drive browsers from other providers: Browserbase, Steel, Anchor and Browser Use. A successful run can become a playbook that replays with new inputs, and the generated Playwright module is available to inspect. You deploy your own instance with the installer or with templates for Docker, Amazon ECS, Kubernetes and GKE. That edition includes host egress rules and a tamper-evident audit trail.
Who it is for. For teams that automate portals and browser workflows through an agent and want to watch its actions and step in by hand.
Good fit when
- The agent needs a browser with logins already completed for work portals, such as insurer or client accounts
- Portal steps repeat and a successful run should become a playbook with new inputs
- You need to watch the agent and take over when needed by switching to Human Mode
- You want cloud browsers from several providers without changing SDK calls
- A team needs its own deployed instance with egress rules and an audit trail
Not a fit when
- A headless browser for public pages is enough: a light CLI such as agent-browser will do
- You need a fully free license for any commercial use: the SDK and CLI are MIT, the rest is under the Sustainable Use License
- You cannot keep work logins in a browser an agent controls
- You need the HIPAA-ready enterprise edition: it lives in a private repository, and this one is the community edition
Example request
Open the supplier portal in Oya, find the request by the number from the email and export its status, and stop at the sign-in step to hand control to meMCP
- Transport
- http
- Authentication
- API key
Environment variables
OYA_API_KEYrequired, secret- API key created on oyabrowser.com, sent in the Authorization header
Limitations
The MCP server and SDK need an API key, which you create on oyabrowser.com. The SDK requires Node.js 20 or newer. The `@oya-ai/browser` package and the CLI are MIT-licensed, and the rest of the project is available under the Sustainable Use License. The terms for self-hosting and commercial use are in LICENSE.md.
The enterprise edition, including the HIPAA-ready offering, is maintained in a separate private repository and is not available here.
Whether Oya Cloud is reachable from Russia has not been checked.
How to disable. Remove the server from Claude Code with `claude mcp remove oya` and revoke the API key on oyabrowser.com. The desktop app is uninstalled the usual way for your system, and a self-hosted instance is stopped with Docker or your orchestrator.
FAQ
How do I connect Oya Browser to Claude Code?
Create an API key on oyabrowser.com and run `claude mcp add --transport http oya https://oyabrowser.com/mcp/pool --header "Authorization: Bearer $OYA_API_KEY"`. The other route is the desktop app, which connects the assistant from its own interface.
Which browsers and providers are supported?
The SDK and CLI work with Oya Cloud, Browserbase, Steel, Anchor, Browser Use and your own browser. `npx @oya-ai/cli init` configures the provider, and SDK calls stay the same when you switch. Logins can be imported from Chrome, Arc, Brave, Edge and Firefox.
What is a playbook?
It is a successful run saved as a repeatable task. You run it again with new inputs, and playbooks can be exported and imported. Details are in the portal automation section of the SDK docs.
Can I deploy Oya myself?
Yes. The installer runs with `curl -fsSL https://raw.githubusercontent.com/OyadotAI/oya-browser/main/install.sh | sh`, and production templates exist for Docker, Amazon ECS, Kubernetes and GKE. Then use your instance's address instead of the cloud MCP address.
Where does an agent find the tool reference?
Oya publishes it at https://oyabrowser.com/llms.txt. The agent only needs to read that file before starting.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server that lets an agent drive a real browser: open pages, click and fill forms through Playwright
A Claude Code plugin with an MCP server and seven skills: the agent drives a real browser by writing Python code
A separate browser for agents with a skill and MCP: Claude Code and Codex work in your logged-in accounts, and every session is recorded for replay
A CLI and MCP server that lets an agent drive your real Chrome with its logged-in sessions: 103 commands across 36 platforms plus browser automation
An MCP server and Chrome extension that let the agent drive your already open browser with your logged-in sessions, all locally
An MCP bridge to control your browser from a coding agent: a Chrome extension and a set of tab and page automation tools
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Oya Browser (this entry) | yes | API key | 351 | |
| agent-browser | yes | not required | 44,858 | a fast Rust CLI for browser automation with accessibility snapshots |
| Browser Use | no | 119,502 | a CLI, skill and Python library for driving a browser | |
| Browser MCP | yes | not required | 7,246 | a Chrome extension and MCP for your already open browser |
README in short
The description opens with a tagline about a browser for regulated industries and splits getting started into three routes: with an assistant, from Node.js and over MCP. A separate block addresses agents themselves and links to a reference, followed by self-hosting and a table of the repository's packages.