Building AI agents, atomically
Example Projects
This section contains detailed examples of using Atomic Agents in various scenarios.
All examples are available in optimized formats for AI assistants:
- **{download}`Examples with documentation <../_static/llms-examples.txt>`** - All examples with source code and READMEs
- **{download}`Full framework package <../_static/llms-full.txt>`** - Complete documentation, source, and examples
Quickstart Examples
Simple examples to get started with the framework:
- Basic chatbot with history
- Custom chatbot with personality
- Streaming responses
- Custom input/output schemas
- Multiple provider support
π View on GitHub - Browse the complete source code and run the examples
Hooks System
Comprehensive monitoring and error handling with the AtomicAgent hook system:
- Parse error handling and validation
- API call monitoring and metrics
- Response time tracking and performance analysis
- Intelligent retry mechanisms
- Production-ready error isolation
- Real-time performance dashboards
π View on GitHub - Browse the complete source code and run the examples
Phoenix Tracing
Trace agents and tools in Arize Phoenix with the optional OpenInference instrumentor:
- Automatic AGENT and TOOL spans via
phoenix.otel.register(auto_instrument=True) - LLM spans from the OpenAI instrumentor nested under each agent run
- A parent chain span that ties a planner agent, a calculator tool and an answer agent into one trace
- Session grouping with
using_session
π View on GitHub - Browse the complete source code and run the examples
Basic Multimodal
Examples of working with images and text:
- Image analysis with text descriptions
- Image-based question answering
- Visual content generation
- Multi-image comparisons
π View on GitHub - Browse the complete source code and run the examples
RAG Chatbot
Build context-aware chatbots with retrieval-augmented generation:
- Document indexing and embedding
- Semantic search integration
- Context-aware responses
- Source attribution
- Follow-up suggestions
π View on GitHub - Browse the complete source code and run the examples
Web Search Agent
Create agents that can search and analyze web content:
- Web search integration
- Content extraction
- Result synthesis
- Multi-source research
- Citation tracking
π View on GitHub - Browse the complete source code and run the examples
Deep Research
Perform comprehensive research tasks:
- Multi-step research workflows
- Information synthesis
- Source validation
- Structured output generation
- Citation management
π View on GitHub - Browse the complete source code and run the examples
YouTube Summarizer
Extract and analyze information from videos:
- Transcript extraction
- Content summarization
- Key point identification
- Timestamp linking
- Chapter generation
π View on GitHub - Browse the complete source code and run the examples
YouTube to Recipe
Convert cooking videos into structured recipes:
- Video analysis
- Recipe extraction
- Ingredient parsing
- Step-by-step instructions
- Time and temperature conversion
π View on GitHub - Browse the complete source code and run the examples
Orchestration Agent
Coordinate multiple agents for complex tasks:
- Agent coordination
- Task decomposition
- Progress tracking
- Error handling
- Result aggregation
π View on GitHub - Browse the complete source code and run the examples
MCP Agent
Build intelligent agents using the Model Context Protocol:
- Server implementation with multiple transport methods
- Dynamic tool discovery and registration
- Natural language query processing
- Stateful conversation handling
- Extensible tool architecture
View MCP Agent Documentation π View on GitHub - Browse the complete source code and run the examples