ToolExperimentalMCP

browser-operator-core

Name: browser-operator-core
Rating: 3.0 (497 reviews)
Author: BrowserOperator

by BrowserOperator

AI browser that runs multi-agent workflows and logs agent interactions

TypeScript

Updated Mar 19, 2026

497

Stars

Forks

View on GitHub

How It Works

Provides an AI-powered browser that runs multi-agent workflows inside the browser environment. Combines a multi-agent runtime, task routing, and UI integration so agents can browse, fetch, and act on web content while coordinating with specialist agents. Distinctive features include built-in agent workflows, MCP support for agent-to-agent messaging, and TypeScript SDKs for extending browser agents. MCP support for agent-to-agent messaging and TypeScript SDKs for extending browser agents.

Why It Matters

As agents act on behalf of users in web contexts, being able to observe and evaluate agent behavior during real interactions becomes essential. BrowserOperator surfaces agent actions, delegation chains, and interaction logs that help forensic evaluation and continuous assessment of agent reliability. That visibility makes it easier to collect agent track records and signal when an agent's web behavior needs governance or retraining. Evaluation-Driven Development (EDDOps).

Target Use Cases

Developers building agentic web experiences who need local multi-agent orchestration, browsing automation, and interaction traces for evaluation. Supervisor Pattern and interaction traces

How It's Used

Automating complex web tasks by delegating subtasks to specialist agents in-browser
Recording agent decision trails and web actions for post-hoc evaluation and debugging
Prototyping agent-to-agent interaction patterns using MCP messaging inside a browser runtime

See related protocols

Standards this tool supports