Shell Browser Agent
shellbrowser.com
- Free plan
- Windows
- macOS
- Self-hosted
What Shell Browser Agent is
Shell Browser Agent is a desktop AI agent for web tasks and browser automation inside authorized profiles. It can navigate, click, type, select, scroll, and drag on pages, read content, capture screenshots, upload files, and verify results. From a task goal, it can create or update an automation, select profiles, and run a guided test. Approved local commands can process CSV, JSON, images, and result files. The agent connects team Streamable HTTP services through MCP with credentials and access tokens, while Local Skills provide team knowledge and procedures. Teams can control access, headless mode, and confirmation behavior for local commands. It supports Windows 10 or 11 on 64-bit devices and macOS 12 or later on Apple silicon and Intel. Plans range from a free option to paid Basic, Professional, Business, and Enterprise tiers, with Enterprise offering private deployment.
Who it suits
It suits business and professional teams automating browser work across authorized profiles. Teams needing local commands, shared services, or private deployment can choose controls and plans accordingly.
Good to know, and what to watch
- Handles broad browser interactions
- Creates and tests automations from goals
- Processes local CSV and JSON files
- Supports team MCP services
- Enterprise private deployment available
- Uses a connected third-party AI key
- AI provider records token use
- Paid plans charge per window and member
- No system is perfectly secure
Plans
Every plan and priceBasic
Business
Enterprise
Professional
Free plan
The verdict
Shell Browser Agent combines browser control, workflow creation, local processing, and team integrations on desktop. Plan selection depends on windows, members, controls, and whether private deployment is required.
What the maker publishes
- Agent permissions
- Teams can enable or disable agent access, configure headless mode, and set confirmation behavior for local commands. source
- AI key and model cost
- Using AI features requires connecting a third-party AI key, and its token usage is recorded with that provider account. source
- AI provider costs
- The agent uses a connected third-party AI key, and the AI provider records token use in that account. source
- Automation
- Shell Browser offers visual workflow nodes, batch runs, logs, and local schedules for recurring tasks. source
- Browser actions
- It can navigate, click, type, select, scroll, and drag on pages, including across authorized profiles. source
- Company facts
- The terms identify the service provider only as the Shell Browser operator named in the product, order, invoice, or transaction record and do not state a legal entity name, headquarters, or founding year. source
- Data handling
- The privacy policy says some profile and task data is stored locally and is sent when users enable sync, sharing, backup, support, or another connected feature. source
- Documentation
- The help center groups product guides across getting started, profiles, fingerprints, proxies, automation, extensions, AI Agent, and plans, with setup and troubleshooting resources. source
- Enterprise option
- Enterprise supports dedicated networks, data retention policies, and private deployment, with architecture and scope confirmed by an advisor. source
- Fingerprint controls
- The product describes independent fingerprint seeds and configurable system, browser, language, hardware, and signal-protection settings for profiles. source
- Independent app listings
- No attributable Apple App Store, Google Play, Chrome Web Store, or Microsoft Store listing for Shell Browser Agent was verified in the pages reviewed. source
- Independent review coverage
- Searches of major review platforms did not surface an attributable Shell Browser Agent customer rating or verified review count; accessible results included unrelated products with similar names. source
- Integrations
- The agent connects team Streamable HTTP services through MCP with required credentials and access tokens. source
- Intended users
- The terms describe the services as built for business and professional use. source
- License
- The terms grant a limited, non-exclusive, non-transferable, non-sublicensable, revocable right to use the service for lawful internal business; the product is not presented as open source. source
- Limit: account guarantees
- The terms say profile isolation, proxies, sync, and automation cannot guarantee freedom from third-party account linkage, limits, or suspension. source
- Limit: Chromium engine
- The product page describes browser versions paired with its current Chromium kernel and says target-site behavior and resource use still need real-world testing. source
- Limit: operating systems
- The download page lists desktop installers for Windows and macOS; it does not list Linux, iOS, Android, or browser-extension apps. source
- Limit: service availability
- The terms say the service is provided as available and does not promise uninterrupted service, full security, or support for every website. source
- Local commands
- Approved local commands can process CSV, JSON, images, and result files. source
- Local commands
- With approval, local commands can process CSV, JSON, image, and task-result files. source
- MCP connections
- Teams can connect Streamable HTTP services and provide required credentials and access tokens to extend agent capabilities. source
- Model providers
- Team settings allow a team to configure its model provider, agent access, headless mode, and approval rules for local commands. source
- Multi-profile tasks
- It can switch among authorized profiles and handle tabs, popups, waits, and tasks spanning multiple pages. source
Checked against the maker's pages on 21 Oct 2026.