Examples

Learn by example with practical AgentiCraft demonstrations.

Quick Start Examples

Hello World

The simplest possible agent - perfect for getting started.

Basic Chat

Build a conversational AI in minutes.

Feature Showcases

Provider Switching

• Runtime provider changes
• Cost optimization strategies
• Automatic failover

Advanced Agents

• ReasoningAgent with transparent thinking
• WorkflowAgent for complex processes
• Combining agent types

Reasoning Patterns

• Chain of Thought for step-by-step analysis
• Tree of Thoughts for exploring alternatives
• ReAct for tool-based reasoning

Real-World Applications

Customer Support Bot

Multi-provider support agent with knowledge base integration.

Data Analysis Pipeline

Workflow agent that processes data through multiple stages.

Content Generator

ReasoningAgent that creates high-quality content with citations.

Code Snippets

Dynamic Model Selection

# Use expensive model for complex tasks
if task.complexity > 0.7:
    agent.set_provider("anthropic", model="claude-3-opus-20240229")
else:
    agent.set_provider("ollama", model="llama2")

Error Recovery

try:
    response = agent.run(prompt)
except ProviderError:
    # Automatic failover
    agent.set_provider("ollama", model="llama2")
    response = agent.run(prompt)

Tool Integration

@tool
def search(query: str) -> str:
    """Search the web."""
    # Implementation

agent = Agent("SearchBot", tools=[search])

Running the Examples

1. Clone the repository:

   git clone https://github.com/agenticraft/agenticraft
   cd agenticraft/examples
   

2. Install dependencies:

   pip install agenticraft
   

3. Set up API keys:

   export OPENAI_API_KEY="your-key"
   export ANTHROPIC_API_KEY="your-key"
   

4. Run examples:

   python hello_world.py
   python provider_switching/basic.py
   

Reasoning Pattern Examples

Chain of Thought

from agenticraft.agents.reasoning import ReasoningAgent

agent = ReasoningAgent(
    name="Analyst",
    reasoning_pattern="chain_of_thought"
)

response = await agent.think_and_act(
    "Calculate the ROI of solar panels over 10 years"
)

# See step-by-step reasoning
for step in response.reasoning_steps:
    print(f"{step.number}. {step.description} ({step.confidence:.0%})")

Tree of Thoughts

agent = ReasoningAgent(
    name="Designer",
    reasoning_pattern="tree_of_thoughts",
    pattern_config={"beam_width": 4}
)

response = await agent.think_and_act(
    "Design a user-friendly mobile app for seniors"
)

# Visualize exploration tree
print(agent.advanced_reasoning.visualize_tree())

ReAct Pattern

from agenticraft.tools import SearchTool, CalculatorTool

agent = ReasoningAgent(
    name="Researcher",
    reasoning_pattern="react",
    tools=[SearchTool(), CalculatorTool()]
)

response = await agent.think_and_act(
    "What's the current GDP per capita of Japan in USD?"
)

# See thought-action-observation cycles
for step in response.reasoning_steps:
    if step.tool_used:
        print(f"Used {step.tool_used}: {step.tool_input}")

Pattern Comparison

# Compare patterns on the same problem
patterns = ["chain_of_thought", "tree_of_thoughts", "react"]
results = {}

for pattern in patterns:
    agent = ReasoningAgent(reasoning_pattern=pattern)
    response = await agent.think_and_act("Solve: 2x + 5 = 15")
    results[pattern] = {
        "answer": response.content,
        "steps": len(response.reasoning_steps),
        "confidence": response.confidence
    }

# Analyze which pattern worked best
for pattern, result in results.items():
    print(f"{pattern}: {result['steps']} steps, {result['confidence']:.0%} confidence")

Contributing Examples

Have a cool use case? We'd love to see it! Share your examples on GitHub.