Agentic Browser

AI agent for web automation and scraping, designed to plan, execute, and verify actions across any website

Source 14 Live site

INDEX

Index for Agentic Browser docs. Topical folders are the primary layout; older numbered guides are folded into the same tree.

Quick navigation#

Core documentation#

  • [Project Overview](Project%20Overview .md) - System architecture, core components, and capabilities
  • [Getting Started](Getting%20Started .md) - Installation, setup, and quick start guide

System components#

Backend architecture#

  • API Server - FastAPI router-service-tool architecture
  • API endpoints and route definitions
  • Service layer business logic
  • Tool layer for external integrations

Agent system#

  • AI Agent System - LangGraph-based agent orchestration
  • React agent for conversational AI
  • Browser Use agent for script generation
  • Tool system and orchestration
  • Context management and state handling

Browser extension#

  • Browser Extension - WebExtensions-based UI and automation
  • Extension architecture and messaging
  • Background and content scripts
  • Side panel UI components
  • Agent execution engine
  • WebSocket communication
  • Authentication system

MCP server#

  • [MCP Server](MCP%20Server .md) - Model Context Protocol implementation
  • Tool definitions and standardization
  • LLM integration through MCP
  • Integration with external clients

Features & integrations#

Service integrations#

  • Service Integrations - External service connectivity
  • Gmail integration for email operations
  • Google Calendar integration for scheduling
  • GitHub integration for repository analysis
  • YouTube integration for video processing
  • Website analysis and content extraction
  • Academic portal integration

Data & models#

  • Data Models and Schemas - API contracts and schemas
  • Request and response models
  • Service integration data structures
  • Agent communication formats
  • Validation and error handling

Configuration & operations#

  • [Configuration Management](Configuration%20Management .md) - Environment setup and configuration
  • [Deployment and Operations](Deployment%20and%20Operations .md) - Production deployment
  • [Development Guidelines](Development%20Guidelines .md) - Contribution and development practices
  • [Testing Strategy](Testing%20Strategy .md) - Testing approach and utilities

Reference#

  • System Architecture - Detailed system design
  • Tool System - Tool definitions and framework
  • Prompts and Prompt Engineering - LLM prompt strategies
  • [Security Considerations](Security%20Considerations .md) - Threat model, approvals, and key handling
  • [Troubleshooting and FAQ](Troubleshooting%20and%20FAQ .md) - Common issues and solutions

Documentation structure#

Two ways to browse the same material:

Topical organization (primary)#

Primary layout is by topic/component:

  • Overview and getting started materials at the top level
  • Component-specific sections in dedicated folders
  • Detailed guides and reference material within each component

Historical reference (numbered sections)#

The original numbered documentation structure is preserved for reference:

  • Section 1: Overview and system architecture
  • Section 2: Installation and getting started
  • Section 3: Python backend API
  • Section 4: Agent intelligence system
  • Section 5: Browser extension
  • Section 6: Data models and API contracts

Both structures provide the same information, choose whichever navigation style works best for you.

Getting started paths#

For first-time users#

  1. Start with [Project Overview](Project%20Overview .md) to understand the system
  2. Follow [Getting Started](Getting%20Started .md) for installation and setup
  3. Explore System Architecture to understand how components interact

For developers#

  1. Review [Development Guidelines](Development%20Guidelines .md) for contribution process
  2. Study relevant component documentation:
  1. Reference Data Models and Schemas for API contracts

For operators#

  1. Read [Configuration Management](Configuration%20Management .md) to set up your environment
  2. Review [Deployment and Operations](Deployment%20and%20Operations .md) for production setup
  3. Use [Troubleshooting and FAQ](Troubleshooting%20and%20FAQ .md) for common issues

For integration#

  1. Review [MCP Server](MCP%20Server .md) for protocol-level integration
  2. Study Service Integrations for available features
  3. Check Data Models and Schemas for API contracts

Key concepts#

Model-Agnostic design#

The system supports multiple LLM providers (Google Gemini, OpenAI, Anthropic, Ollama, DeepSeek, OpenRouter) through a unified abstraction layer. Users supply their own API keys via environment variables (BYOK - Bring Your Own Keys).

Stack#

  • Frontend: Browser extension with background and content scripts
  • Backend: FastAPI server with service-oriented architecture
  • Agent Runtime: LangGraph-based agent orchestration
  • LLM Layer: Model-agnostic provider adapters
  • Safety Layer: Guardrails, logging, and user consent

Tool system#

The system provides 11+ specialized tools for web automation, content processing, and external service integration. Tools are dynamically constructed based on context and user authentication.

Declarative action system#

Browser automation uses JSON action plans from the LLM. Plans are visible and require approval before they run.

Technology stack#

  • Language: Python 3.12+ (backend), TypeScript/React (frontend)
  • Agent Framework: LangChain, LangGraph
  • Web Framework: FastAPI, Uvicorn
  • Extension Framework: WXT (Web eXtension Tooling)
  • Browser APIs: WebExtensions API
  • Communication: MCP (Model Context Protocol), WebSocket, HTTP/REST
  • Content Processing: BeautifulSoup, html2text, yt-dlp
  • Security: python-dotenv, pycryptodome

Common tasks#

Setting up the development environment#

See: [Getting Started](Getting%20Started .md#installation) and [Development Guidelines](Development%20Guidelines .md)

Configuring API keys and environment#

See: [Configuration Management](Configuration%20Management .md)

Adding a new service integration#

See: Service Integrations and [Development Guidelines](Development%20Guidelines .md)

Understanding agent behavior#

See: AI Agent System and Prompts and Prompt Engineering

Debugging and troubleshooting#

See: [Troubleshooting and FAQ](Troubleshooting%20and%20FAQ .md) and System Architecture

Deploying to production#

See: [Deployment and Operations](Deployment%20and%20Operations .md)

Integrating with external tools#

See: [MCP Server](MCP%20Server .md) and Data Models and Schemas

Contributing#

Before contributing, please review:

  1. [Development Guidelines](Development%20Guidelines .md) - Contribution process and standards
  2. [Testing Strategy](Testing%20Strategy .md) - Testing requirements
  3. [Security Considerations](Security%20Considerations .md) - Threat model, approvals, and key handling

Version history#

This consolidated documentation combines:

  • New Documentation: Topical organization with component-based structure
  • Previous Documentation: Detailed numbered guides with implementation specifics

Both sources have been merged to provide detailed coverage of all aspects of the Agentic Browser system.

Additional resources#

Last Updated: 2026