Agent-E
VerifiedOpen-source agent for natural language browser automation and task handling.
What is Agent-E?
Agent-E is an open-source agent framework built on AG2 that focuses on automating everyday browser interactions. Users issue natural language commands to complete actions such as filling web forms, sorting e-commerce results, locating specific page details, or managing project tickets.
The system works by translating instructions into browser actions via Playwright or local Chrome, with support for multi-agent orchestration in its managed version. Configuration happens through environment files and LLM settings, allowing flexibility across OpenAI and other models.
It suits developers and power users who want to script repetitive web tasks without writing custom code for each site. The project remains under active development with emphasis on practical browser workflows rather than desktop or PDF automation.
Capabilities
What you can build with Agent-E
E-commerce product research
Search and filter items on sites like Amazon by price, ratings, or best-seller status using simple prompts.
Web content lookup
Find sports scores, contact details, or historical information across multiple pages without manual navigation.
Project workflow management
Filter and organize issues in JIRA or similar platforms to streamline daily task handling.
Install Agent-E
python -m ae.main./install.sh- 1Run the provided install script for your OS (install.sh on macOS/Linux or win_install.ps1 on Windows) with optional -p flag for Playwright.
- 2Create and activate a Python 3.11 virtual environment using uv.
- 3Compile dependencies from pyproject.toml and install them with uv pip.
- 4Copy .env-example to .env and add your LLM model name plus API key.
- 5Launch the agent with python -m ae.main (or python -u -m ae.main on macOS).
Works with
Agent-E: pros & cons
Pros
- +Natural language control reduces need for site-specific scripts
- +Open-source with clear installation paths for quick local testing
- +Covers a wide range of common browser tasks out of the box
- +Supports both local Chrome and Playwright for flexibility
Cons
- –Currently limited to browser automation only
- –Requires manual environment and API key setup
- –Performance depends on chosen LLM quality and speed
Frequently asked questions
It is built on the AG2 agent framework for handling multi-agent interactions and browser control.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other web & browser options worth comparing.
