Install

Agent-E requires Python 3.11 (3.10+ also works) and uses uv for environment and dependency management. You also need a browser: either local Google Chrome or Playwright.

Prerequisites

• Python 3.11+
• uv package manager
• An LLM API key (OpenAI gpt-4-turbo recommended; Azure OpenAI and other OpenAI-compatible endpoints are supported)
• Google Chrome installed locally, OR Playwright (installed in step 4 below)

Install uv if you do not have it:

macOS / Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows:

powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

Alternatively: pip install uv

---

Quick install (recommended)

Clone the repo, then run the provided install script.

macOS / Linux:

git clone https://github.com/EmergenceAI/Agent-E.git
cd Agent-E
./install.sh

Add -p to also install Playwright browser drivers without an interactive prompt:

./install.sh -p

Windows (PowerShell):

git clone https://github.com/EmergenceAI/Agent-E.git
cd Agent-E
.\win_install.ps1

Add -p for Playwright support:

.\win_install.ps1 -p

---

Manual install

If you prefer to set up the environment step by step:

# 1. Create and activate a virtual environment
uv venv --python 3.11
source .venv/bin/activate        # Windows: .venv\Scripts\activate

# 2. Generate requirements.txt and install dependencies
uv pip compile pyproject.toml -o requirements.txt
uv pip install -r requirements.txt

# 3. (Optional) Install dev extras
uv pip install -r pyproject.toml --extra dev

# 4. (Optional) Install Playwright drivers if not using local Chrome
playwright install

---

Configure environment variables

Copy the example env file and fill in the required values:

cp .env-example .env

Minimum required variables in .env:

AUTOGEN_MODEL_NAME=gpt-4-turbo
AUTOGEN_MODEL_API_KEY=<your-openai-api-key>

For Azure OpenAI or other compatible endpoints, also set:

AUTOGEN_MODEL_BASE_URL=https://<YOUR_AZURE_ENDPOINT>.openai.azure.com
AUTOGEN_MODEL_API_TYPE=azure
AUTOGEN_MODEL_API_VERSION=2023-03-15-preview

Optional variables:

Variable	Default	Description
AUTOGEN_LLM_TEMPERATURE	0.0	Controls output randomness
AUTOGEN_LLM_TOP_P	0.001	Token sampling diversity
BROWSER_STORAGE_DIR	(none)	Path to your local Chrome profile (from chrome://version/)
SAVE_CHAT_LOGS_TO_FILE	true	Save chat logs to file vs stdout
LOG_MESSAGES_FORMAT	text	json or text
ADDITIONAL_SKILL_DIRS	(none)	Comma-separated dirs or .py files for custom skills
PLANNER_USER_INPUT_SKILL_ENABLED	false	Allow planner to request user input

Also copy and edit the LLM config:

cp agents_llm_config-example.json agents_llm_config.json
# Edit agents_llm_config.json to match your model settings