designing-real-world-ai-agents-workshop vs Browser Use

Executes multi-turn research workflows using Google Gemini API with built-in Google Search grounding to retrieve factual, up-to-date information. The Deep Research Agent (src/research/server.py) implements a tool-use pattern where Gemini can invoke search tools iteratively, refining queries based on intermediate results, and persists findings into a structured research.md file. Supports YouTube transcript extraction when URLs are provided, enabling multi-modal source integration.

Unique: Uses Gemini's native Google Search grounding (not external RAG) combined with tool-use agents for iterative query refinement, eliminating hallucination risk while maintaining real-time information access. YouTube transcript extraction is built-in, enabling multi-modal research without separate API calls.

vs alternatives: Faster and more accurate than RAG-based research systems because it queries live search results directly rather than relying on static embeddings, and cheaper than multi-step LLM chains because grounding is native to Gemini's API.

Implements a two-server MCP architecture (Deep Research Agent + LinkedIn Writer Agent) using FastMCP framework, where each server exposes tools, resources, and prompts independently and communicates through standardized MCP protocol. The architecture decouples research and writing concerns, allowing each agent to be developed, tested, and scaled independently while maintaining a unified interface. Configuration is managed via .mcp.json and environment variables, enabling runtime server discovery and tool registration.

Unique: Uses FastMCP framework to expose agents as standardized MCP servers rather than monolithic functions, enabling true decoupling where each agent (research, writing) has its own process, configuration, and tool registry. This pattern allows IDE integration (Claude Code, Cursor) without custom client code.

vs alternatives: More modular and testable than LangChain agent chains because each agent is independently deployable and has explicit tool/resource contracts, and more flexible than REST-based agent APIs because MCP provides native IDE integration without custom UI.

Centralizes configuration using Pydantic Settings models (src/research/config/, src/writing/config/) that load from environment variables and .env files, enabling environment-specific configuration without code changes. Configuration includes API keys, model parameters, evaluation thresholds, and server endpoints. Pydantic validation ensures type safety and provides helpful error messages for missing or invalid configuration.

Unique: Uses Pydantic Settings for type-safe, validated configuration with automatic environment variable loading. Configuration is centralized in dedicated config modules (src/research/config/, src/writing/config/), making it easy to add new configuration options without modifying agent code.

vs alternatives: More robust than manual environment variable parsing because Pydantic validates types and provides helpful error messages, and more maintainable than hardcoded configuration because all settings are in one place.

Persists research findings to a structured markdown file (research.md) that serves as the knowledge base for the writing agent. The markdown format enables human readability while maintaining machine-parseable structure (headings, lists, citations). Research findings include source citations, timestamps, and iterative search history, creating an auditable record of how conclusions were reached. The writing agent reads this markdown to generate content, ensuring factual grounding.

Unique: Uses markdown as the primary knowledge representation format, enabling both machine parsing (for writing agent) and human inspection (for manual review). Includes source citations and search history, creating an auditable record of research methodology.

vs alternatives: More transparent than vector databases because research is human-readable and manually editable, and more flexible than structured databases because markdown can accommodate unstructured notes and citations.

Implements a multi-iteration content generation and evaluation pattern in the LinkedIn Writer Agent (src/writing/server.py) where an LLM generates initial content, an evaluator (LLM-as-judge) scores it against quality criteria, and an optimizer refines it based on feedback. The loop continues until quality thresholds are met or max iterations reached. Uses Opik for tracing and LLM-based evaluation metrics, enabling observable, measurable content quality improvement without human-in-the-loop.

Unique: Combines LLM-as-judge evaluation with iterative optimization in a closed loop, using Opik for full observability of each refinement cycle. Unlike simple prompt engineering, this pattern measures quality objectively and refines based on measurable feedback, not heuristics.

vs alternatives: More reliable than single-pass LLM generation because it validates and refines output against explicit criteria, and more transparent than black-box content APIs because every iteration is traced and evaluated metrics are visible.

Integrates Google Gemini's Imagen model for AI-generated images within the writing workflow, enabling automatic image creation to accompany generated LinkedIn posts. The image generation is triggered based on post content and writing profiles, with generated images persisted to the dataset directory. Supports prompt engineering for image generation based on post themes and audience preferences.

Unique: Integrates Imagen directly into the writing workflow as a native step, not a separate tool — image generation is triggered automatically based on post content and writing profiles, enabling end-to-end content creation without manual image selection.

vs alternatives: More integrated than using external image APIs (DALL-E, Midjourney) because it's part of the same Gemini API ecosystem and can reference post content directly, and faster than manual image selection because generation is automated and parallelizable.

Implements a structured dataset system (datasets/ directory) with batch evaluation scripts that process multiple content samples through the writing workflow and score them using LLM-as-judge metrics via Opik. The evaluation system measures quality across dimensions (clarity, engagement, relevance) and aggregates results for statistical analysis. Supports dataset versioning and comparison across model versions or writing profiles.

Unique: Combines structured dataset management with Opik-based LLM-as-judge evaluation, enabling systematic quality measurement across multiple samples with full traceability. Unlike ad-hoc evaluation, this pattern produces reproducible, comparable metrics across writing profiles and model versions.

vs alternatives: More rigorous than manual spot-checking because it evaluates entire datasets systematically, and more transparent than black-box quality scores because each evaluation is traced in Opik with full iteration history visible.

Defines MCP tools and resources using FastMCP decorators (@mcp.tool, @mcp.resource) with JSON schema validation, enabling type-safe tool invocation and automatic schema generation. The research and writing servers expose distinct tool sets (search, research persistence, content generation, evaluation) with Pydantic-based input/output validation. MCP routers (src/research/routers/, src/writing/routers/) map tool invocations to application logic, decoupling tool definitions from implementation.

Unique: Uses FastMCP decorators with Pydantic models to automatically generate MCP tool schemas, eliminating manual JSON schema writing. Router pattern (src/research/routers/, src/writing/routers/) decouples tool definitions from implementation, enabling easy tool addition without modifying server core.

vs alternatives: More maintainable than hand-written JSON schemas because Pydantic models are single source of truth, and more discoverable than REST APIs because MCP clients can introspect tool schemas at runtime without documentation.

browser-use/browser-use | DeepWiki Loading... Index your code with Devin DeepWiki DeepWiki browser-use/browser-use Index your code with Devin Edit Wiki Share Loading... Last indexed: 17 May 2026 ( 933e28 ) Overview System Architecture Installation and Setup Quick Start Examples Agent System Agent Core and Execution Loop Message Manager and Prompt Construction Agent State and History Management System Prompts and Output Formats Skills Integration Agent Configuration and Settings Loop Detection and Behavioral Nudges Message Compaction System Memory and Follow-up Tasks Judge System and Trace Evaluation Browser Session Management BrowserSession Lifecycle Browser Profile Configuration SessionManager and CDP Session Pool Target and Frame Management Navigation and Tab Control Event-Driven Architecture Event System Overview Event Types Reference Watchdog Pattern and Base Classes Core Watchdog Implementations DOM Processing Engine DOM Tree Construction DOM Serialization Pipeline Interactive Element Detection Visibility Calculation and Coordinate Transformation Screenshot Highlighting System Browser State Summary Markdown Extraction and HTML Serialization Tools and Action System Tools Registry and Action Models Built-in Actions Reference Action Execution Pipeline Custom Tools and Extensions Click Action Deep Dive Input Action and Autocomplete Detection FileSystem Integration Br

System Architecture | browser-use/browser-use | DeepWiki Loading... Index your code with Devin DeepWiki DeepWiki browser-use/browser-use Index your code with Devin Edit Wiki Share Loading... Last indexed: 17 May 2026 ( 933e28 ) Overview System Architecture Installation and Setup Quick Start Examples Agent System Agent Core and Execution Loop Message Manager and Prompt Construction Agent State and History Management System Prompts and Output Formats Skills Integration Agent Configuration and Settings Loop Detection and Behavioral Nudges Message Compaction System Memory and Follow-up Tasks Judge System and Trace Evaluation Browser Session Management BrowserSession Lifecycle Browser Profile Configuration SessionManager and CDP Session Pool Target and Frame Management Navigation and Tab Control Event-Driven Architecture Event System Overview Event Types Reference Watchdog Pattern and Base Classes Core Watchdog Implementations DOM Processing Engine DOM Tree Construction DOM Serialization Pipeline Interactive Element Detection Visibility Calculation and Coordinate Transformation Screenshot Highlighting System Browser State Summary Markdown Extraction and HTML Serialization Tools and Action System Tools Registry and Action Models Built-in Actions Reference Action Execution Pipeline Custom Tools and Extensions Click Action Deep Dive Input Action and Autocomplete Detection FileS

Agent System | browser-use/browser-use | DeepWiki Loading... Index your code with Devin DeepWiki DeepWiki browser-use/browser-use Index your code with Devin Edit Wiki Share Loading... Last indexed: 17 May 2026 ( 933e28 ) Overview System Architecture Installation and Setup Quick Start Examples Agent System Agent Core and Execution Loop Message Manager and Prompt Construction Agent State and History Management System Prompts and Output Formats Skills Integration Agent Configuration and Settings Loop Detection and Behavioral Nudges Message Compaction System Memory and Follow-up Tasks Judge System and Trace Evaluation Browser Session Management BrowserSession Lifecycle Browser Profile Configuration SessionManager and CDP Session Pool Target and Frame Management Navigation and Tab Control Event-Driven Architecture Event System Overview Event Types Reference Watchdog Pattern and Base Classes Core Watchdog Implementations DOM Processing Engine DOM Tree Construction DOM Serialization Pipeline Interactive Element Detection Visibility Calculation and Coordinate Transformation Screenshot Highlighting System Browser State Summary Markdown Extraction and HTML Serialization Tools and Action System Tools Registry and Action Models Built-in Actions Reference Action Execution Pipeline Custom Tools and Extensions Click Action Deep Dive Input Action and Autocomplete Detection FileSystem I

browser-use/browser-use | DeepWiki Loading... Index your code with Devin DeepWiki DeepWiki browser-use/browser-use Index your code with Devin Edit Wiki Share Loading... Last indexed: 17 May 2026 ( 933e28 ) Overview System Architecture Installation and Setup Quick Start Examples Agent System Agent Core and Execution Loop Message Manager and Prompt Construction Agent State and History Management System Prompts and Output Formats Skills Integration Agent Configuration and Settings Loop Detection and Behavioral Nudges Message Compaction System Memory and Follow-up Tasks Judge System and Trace Evaluation Browser Session Management BrowserSession Lifecycle Browser Profile Configuration SessionManager and CDP Session Pool Target and Frame Management Navigation and Tab Control Event-Driven Architecture Event System Overview Event Types Reference Watchdog Pattern and Base Classes Core Watchdog Implementations DOM Processing Engine DOM Tree Construction DOM Serialization Pipeline Interactive Element Detection Visibility Calculation and Coordinate Transformation Screenshot Highlighting System Browser Sta