Agent E

Open-source AI agent that automates tasks directly in your local browser

5.0 (4)

Прегледано от Daniel Nikulshyn·Актуализирано май 2026 г.

Web Scraping Research Open Source AI Agents Local-First Developer Tools Browser Automation

Преглед

Agent E is a web automation agent that controls a browser on your own machine to complete multi-step online tasks. It can navigate sites, fill forms, extract information, and chain actions together based on natural language instructions. Built with a hierarchical architecture that separates planning from execution, Agent E aims to handle complex, real-world web workflows more reliably than single-prompt approaches. Because it runs locally, users keep their sessions, cookies, and credentials on their own device rather than routing them through a remote service. It is geared toward developers and researchers exploring agentic browsing, benchmarking web automation, or building custom assistants for repetitive online work.

Ключови функции

Local browser automation
Hierarchical planner and executor agents
Natural language task instructions
Form filling and data extraction
Support for multi-step workflows
Configurable LLM backends

Случаи на употреба

Automate Multi-Step Web Workflows

Use natural language to instruct Agent E to navigate websites, fill forms, and chain actions together for repetitive online tasks without writing custom scripts.

Private Browsing with Logged-In Sessions

Run automation locally so cookies, credentials, and sessions remain on your machine, enabling tasks on authenticated sites without exposing data to third-party services.

Benchmark Agentic Browsing Research

Researchers can evaluate hierarchical planner-executor architectures on web automation tasks and compare performance across different configurable LLM backends.

Extract Structured Data from Websites

Direct the agent to visit pages, gather specific information, and return structured results for analysis, monitoring, or feeding into downstream applications.

Плюсове и минуси

Плюсове

Runs locally using your own browser session
Handles multi-step web tasks via natural language
Open architecture suitable for research and extension
Avoids sending browsing data through third-party services

Минуси

Requires technical setup and command-line familiarity
Reliability varies on complex or dynamic websites
Performance depends on the underlying LLM provider

Отзиви

5.0

Средно от 4 оценки.

Влез, за да оставиш отзив.

Jamal Carter

Years in this space

I've evaluated a lot of these over the years. What stands out here is natural language task instructions — handled better than most — and open architecture suitable for research and extension. Worth the time if this is your use case.

Pierre Dubois

Years in this space

I've evaluated a lot of these over the years. What stands out here is configurable LLM backends — handled better than most — and avoids sending browsing data through third-party services. Requires technical setup and command-line familiarity is my one real gripe. Worth the time if this is your use case.

Kwame Mensah

Compared a few options

Evaluated this against two competitors. Where it wins: local browser automation and open architecture suitable for research and extension. On balance the feature set — especially support for multi-step workflows — justifies the 5 stars for our use case.

Aisha Khan

Compared a few options

Evaluated this against two competitors. Where it wins: configurable LLM backends and open architecture suitable for research and extension. On balance the feature set — especially configurable LLM backends — justifies the 5 stars for our use case.