About
Grasp is a self-hosted, open-source agentic browser designed to seamlessly integrate with AI applications and agents. It allows for human-like browser automation through natural language control, eliminating the need for coding. Dockerized for isolation and easy deployment, Grasp enables agents to interact with web content, log into sites, and perform tasks as a human user would.
Key Features
- Isolated browser environment via Docker
- Human-like browser automation
- Natural language control (no code required)
- Supports multiple providers (Anthropic, OpenAI, Amazon Bedrock, Azure OpenAI)
- Built-in MCP and A2A support
- 39 GitHub stars
Use Cases
- Automated web browsing for AI agents
- Planning trips with AI assistance
- Integrating with Claude Desktop
