> ## Documentation Index
> Fetch the complete documentation index at: https://ekacare-fix-ts-sdk-github-npm-links.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Agents

> Build agents with GenericAgent

## GenericAgent

The main agent class that orchestrates LLM, tools, and conversation:

```python theme={null}
from echo import GenericAgent, AgentConfig, LLMConfig, ConversationContext

agent = GenericAgent(
    agent_config=agent_config,
    llm_config=llm_config,
    tools=[tool1, tool2],  # Optional tools
)

# Run with conversation context
context = ConversationContext()
context.add_message(Message(
    role=MessageRole.USER,
    content=[TextMessage(text="Your query here")],
))

result = await agent.run(context)
print(result.llm_response.text)
```

## Agent Result

The `run()` method returns an `AgentResult`:

```python theme={null}
result = await agent.run(context)

# Response text
print(result.llm_response.text)

# Updated context (for multi-turn)
context = result.context

# Tool calls made
for item in result.llm_response.verbose:
    if item.type == "tool":
        print(f"Called: {item.tool_name}")

# Elicitations (UI components)
for elicit in result.llm_response.elicitations:
    print(f"Component: {elicit.details.component}")

# Errors
if result.error:
    print(f"Error: {result.error}")
```

## Multi-Turn Conversations

Maintain context across turns:

```python theme={null}
context = ConversationContext()

# Turn 1
context.add_message(Message(role=MessageRole.USER, content=[TextMessage(text="What is diabetes?")]))
result = await agent.run(context)
context = result.context

# Turn 2 (agent remembers context)
context.add_message(Message(role=MessageRole.USER, content=[TextMessage(text="What are the types?")]))
result = await agent.run(context)
```

## Handling Elicitations

When agent returns UI components for user input:

```python theme={null}
result = await agent.run(context)

if result.llm_response.elicitations:
    elicit = result.llm_response.elicitations[0]

    # Show options to user
    options = elicit.details.input.get('options', [])
    for i, opt in enumerate(options, 1):
        print(f"{i}. {opt}")

    # Get user selection and add as tool result
    user_choice = get_user_input()
    context.add_message(Message(
        role=MessageRole.TOOL,
        content=[ToolResult(tool_id=elicit.tool_id, result=user_choice)],
    ))
```

## Next Steps

<CardGroup cols={2}>
  <Card title="Tools" icon="wrench" href="/ai-tools/agent-kit/tools">
    Add tools to agents
  </Card>

  <Card title="Streaming" icon="stream" href="/ai-tools/agent-kit/streaming">
    Real-time responses
  </Card>
</CardGroup>