Invisible Playwright MCP — An AI agent that browses the web like a human, bypassing anti-bot detection and captchas.
Analyzed by Sai Pavan Gopularam · AI · Web Automation · View on GitHub
- Stars: 31663
- Forks: 4676
- Commits last 30 days: 100
- Health: Active (100 commits this month)
- Language: Python
- License: MIT
What It Is
Invisible Playwright MCP is a specialized browser automation tool that lets AI agents interact with websites without being detected as bots. Unlike standard automation frameworks that often trigger captchas or blocks, this project uses stealth techniques to mimic human browsing behavior, making it appear as a regular user.
This matters because many valuable web data sources are protected by sophisticated anti-bot systems. Traditional scraping or automation often fails, making certain data inaccessible. This project kills the problem of getting blocked, opening up new possibilities for AI agents to perform complex tasks like market research, lead generation, or competitive analysis on any public website.
License Verdict
AGPL-3.0 License — Significant Restrictions for SaaS — Commercial Use Restricted • Copyleft Applies to SaaS
The README explicitly states that 'Everything distributed before 2 September 2026 was released under AGPL-3.0 and stays under it.' Since we are currently before that date, this project is effectively under AGPL-3.0. If you use this to provide a network service (like a SaaS product), you are generally required to make your entire application's source code available to your users. This severely limits direct commercial SaaS monetization unless you intend for your product to be fully open source. It is more suitable for internal tools or consulting services where the 'user' is the client receiving the service, not interacting directly with the software over a network.
How to Use It
To get started, you'll install a Python package manager (uv) and then use it to run the Invisible Playwright MCP web UI. You'll need an OpenRouter API key to connect to an LLM for agentic browsing.
Prerequisites:
- Python 3.8+
- OpenRouter API Key
- Linux or Windows PowerShell
Estimated setup time: 10 minutes.
curl -LsSf https://astral.sh/uv/install.sh | sh
source $HOME/.local/bin/env
uvx invisible-playwright-mcp ui --openrouter-key sk-or-YOUR_OPENROUTER_KEY
What I'd Build With This
Niche Job Application Filler (Desktop App) (micro-saas)
Build a desktop application that helps job seekers automatically fill out complex application forms on various job boards, even those with strong bot detection. Users would run the app locally, inputting their resume data, and the agent would navigate, input information, and submit applications. The AGPL license means the app's source code would need to be available, but this is less restrictive for a local client-side tool.
Effort: 2 Weeks Build Time · Target: Job Seekers, Recruiters · Pricing: $29/month per user
Competitive Intelligence Data Service (saas)
Offer a service that provides curated, hard-to-get competitive intelligence data (e.g., competitor pricing changes, new product launches, market trends) from websites using advanced anti-bot measures. Instead of exposing the agent directly, you sell the *data reports* or *alerts*. The AGPL license would apply to your internal data extraction system, but not necessarily to the reports you generate and sell, as the end-user is not interacting with the software as a service.
Effort: 1 Month Build Time · Target: E-commerce Businesses, Market Research Firms · Pricing: $199-$499/month for data feeds
Internal Compliance Monitoring & Reporting System (enterprise)
Develop a custom solution for large enterprises (e.g., financial institutions, legal firms) to monitor public-facing websites for regulatory compliance, legal notices, or specific data points. The system would run entirely within the enterprise's infrastructure, using the invisible agent to gather information from various government, industry, or partner sites. Since it's an internal tool, the AGPL implications are less severe, as the 'users' are internal employees or the organization itself.
Effort: 2-3 Months Build Time · Target: Enterprise Legal & Compliance Departments · Pricing: $10,000 - $50,000+ per project
Sai Pavan Gopularam's Take
This project is a game-changer for anyone needing to automate web tasks that typically get blocked by anti-bot measures. The ability to browse undetected opens up a world of data extraction and process automation that was previously impossible or prohibitively expensive. Despite the current AGPL license, I see a clear path to building a specialized data service, charging $500/month for automated, hard-to-get market intelligence.
Watch Out For
- AGPL-3.0 License Restrictions: As detailed above, the AGPL-3.0 license currently applies, which has significant implications for commercial SaaS products. Carefully consider your distribution model.
- Ethical & Legal Use: Automating web browsing, even stealthily, requires adherence to the terms of service of the websites you interact with. Respect rate limits and ensure your activities are legal and ethical.
- OpenRouter Key Dependency: The web UI requires an OpenRouter API key, meaning you're reliant on their service and pricing for the LLM component. Factor this into your operational costs and potential single point of failure.
- Privacy and Data Flow: While the agent runs locally, your conversations and what the agent reads on pages are sent to OpenRouter. Be mindful of sensitive data and ensure compliance with privacy regulations.
I break down trending repos like Invisible Playwright MCP every week — join the newsletter.