CountBot provides a browser automation CLI designed for AI agents. It enables the programmatic execution of web-based tasks, allowing agents to move beyond static data by interacting directly with live websites.
This tool is built for automation developers and testers who need agents to navigate pages, fill out forms, click buttons, and extract real-time data or screenshots from web applications.
Follow the documented setup, then try a first task.
Installs: Complete CountBot source application, including repository workspace skills; the listed skill paths are a bounded selection. · CountBot
Run from: Begin in the desired checkout parent directory; subsequent commands run inside CountBot.
In a terminal, clone the repository and enter its directory.
git clone https://github.com/countbot-ai/CountBot.git cd CountBot
In the same terminal, install the default dependencies.
pip install -r requirements.txt
Start CountBot from the repository directory. Its local web interface opens at http://127.0.0.1:8000 by default.
python start_app.py
Uses news
Use the news skill to find five recent AI news items. Return a concise list with each item's title, source, link, and a short English summary.
Uses web-design
Use the web-design skill to create a responsive single-page product introduction from the product details I provide. Save the HTML locally for review, with a distinctive visual style and clear feature sections. Do not deploy it.
Uses skill-creator
Use the skill-creator skill to help turn my recurring workflow into a first skill draft. Clarify its goal and steps, draft the instructions, and propose three test prompts for review.
Claude
Codex
Cursor
OpenClaw
OpenCode
DeepSeek