Agent Templates Guide¶

Pre-configured agent templates for common use cases. Templates create fully configured agent instances with all necessary resources.

Location: bast_ai_agno.bp.agent_templates

Reference: Agno Agent Documentation

Table of Contents¶

Overview
Quick Start
Template Types
Usage Examples
Best Practices

Overview¶

Agent templates are high-level functions that create fully configured agent instances with: - All necessary resources (models, storage, knowledge, memory) - Optimized configurations for specific use cases - Ready-to-use agent instances

Templates vs Configurations vs Presets¶

Configurations (dto/agent_configs.py): Dictionary configurations
Presets (dto/agent_preset_utils.py): Functions that return configuration dictionaries
Templates (bp/agent_templates.py): Functions that return fully configured Agent instances

When to Use Templates¶

✅ Use templates when: - You want a ready-to-use agent with minimal setup - You need a common pattern (RAG, tools, conversational, reasoning) - You want all resources automatically configured

❌ Don't use templates when: - You need custom resource setup - You want to reuse existing resources - You need fine-grained control over configuration

Quick Start¶

Basic Template Usage¶

from bast_ai_agno.bp.agent_templates import create_rag_agent_template

# Create a RAG agent with all resources
agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="my_rag_agent",
    knowledge_base_name="my_kb"
)

# Use immediately
response = agent.run(
    input="What is the answer?",
    user_id="user123",
    session_id="session1"
)

Template Types¶

1. RAG Agent Template¶

Creates an agent optimized for knowledge base search.

Function: create_rag_agent_template()

Features: - Knowledge base integration - Automatic knowledge search - Memory capabilities (optional) - RAG-optimized instructions

Parameters: - cat_name (str): Catalog name for resources - agent_name (str): Name for the agent - knowledge_base_name (str, optional): Name of knowledge base (defaults to agent_name) - model_id (str): Model ID (default: "gpt-4o-mini") - search_knowledge (bool): Enable knowledge search (default: True) - enable_memory (bool): Enable memory (default: True) - custom_instructions (List[str], optional): Custom instructions

Example:

from bast_ai_agno.bp.agent_templates import create_rag_agent_template

agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="medical_assistant",
    knowledge_base_name="medical_kb",
    search_knowledge=True,
    enable_memory=True
)

2. Tool-Using Agent Template¶

Creates an agent optimized for external tool integration.

Function: create_tool_agent_template()

Features: - Tool integration - Configurable tool call limits - Tool choice strategies - Tool-optimized instructions

Parameters: - cat_name (str): Catalog name for resources - agent_name (str): Name for the agent - tools (List[Any]): List of tool instances - model_id (str): Model ID (default: "gpt-4o-mini") - tool_call_limit (int, optional): Maximum tool calls per run - tool_choice (str): Tool choice strategy ("auto", "required", "none") (default: "auto") - enable_memory (bool): Enable memory (default: False) - custom_instructions (List[str], optional): Custom instructions

Example:

from bast_ai_agno.bp.agent_templates import create_tool_agent_template
from agno.tools.duckduckgo import DuckDuckGoTools

agent = create_tool_agent_template(
    cat_name="myapp",
    agent_name="research_assistant",
    tools=[DuckDuckGoTools(modifier="site:example.com")],
    tool_call_limit=10,
    tool_choice="auto"
)

3. Conversational Agent Template¶

Creates an agent optimized for multi-turn conversations.

Function: create_conversational_agent_template()

Features: - Session management - Conversation history - Memory capabilities - Session summaries - Conversational-optimized instructions

Parameters: - cat_name (str): Catalog name for resources - agent_name (str): Name for the agent - model_id (str): Model ID (default: "gpt-4o-mini") - enable_memory (bool): Enable memory (default: True) - num_history_runs (int): Number of history runs (default: 10) - enable_session_summaries (bool): Enable session summaries (default: True) - custom_instructions (List[str], optional): Custom instructions

Example:

from bast_ai_agno.bp.agent_templates import create_conversational_agent_template

agent = create_conversational_agent_template(
    cat_name="myapp",
    agent_name="chat_assistant",
    enable_memory=True,
    num_history_runs=15
)

4. Reasoning Agent Template¶

Creates an agent optimized for complex problem solving.

Function: create_reasoning_agent_template()

Features: - Reasoning capabilities - Configurable reasoning steps - Memory capabilities (optional) - Reasoning-optimized instructions

Parameters: - cat_name (str): Catalog name for resources - agent_name (str): Name for the agent - model_id (str): Model ID (default: "gpt-4o-mini") - min_steps (int): Minimum reasoning steps (default: 2) - max_steps (int): Maximum reasoning steps (default: 15) - enable_memory (bool): Enable memory (default: True) - custom_instructions (List[str], optional): Custom instructions

Example:

from bast_ai_agno.bp.agent_templates import create_reasoning_agent_template

agent = create_reasoning_agent_template(
    cat_name="myapp",
    agent_name="analyst",
    min_steps=3,
    max_steps=20
)

Usage Examples¶

Example 1: Simple RAG Agent¶

from bast_ai_agno.bp.agent_templates import create_rag_agent_template

# Create RAG agent
agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="doc_assistant",
    knowledge_base_name="docs_kb"
)

# Use for queries
response = agent.run(
    input="What is the installation process?",
    user_id="user123",
    session_id="session1"
)

Example 2: Tool-Enabled Research Agent¶

from bast_ai_agno.bp.agent_templates import create_tool_agent_template
from agno.tools.duckduckgo import DuckDuckGoTools

# Create tool agent
agent = create_tool_agent_template(
    cat_name="myapp",
    agent_name="researcher",
    tools=[DuckDuckGoTools()],
    tool_call_limit=5
)

# Use for research
response = agent.run(
    input="Find the latest developments in quantum computing",
    user_id="user123",
    session_id="session1"
)

Example 3: Customer Support Agent¶

from bast_ai_agno.bp.agent_templates import create_conversational_agent_template

# Create conversational agent
agent = create_conversational_agent_template(
    cat_name="myapp",
    agent_name="support_agent",
    enable_memory=True,
    num_history_runs=20,
    enable_session_summaries=True
)

# Use for customer support
response = agent.run(
    input="I need help with my account",
    user_id="customer123",
    session_id="support_ticket_1"
)

Example 4: Analysis Agent with Reasoning¶

from bast_ai_agno.bp.agent_templates import create_reasoning_agent_template

# Create reasoning agent
agent = create_reasoning_agent_template(
    cat_name="myapp",
    agent_name="analyst",
    min_steps=3,
    max_steps=15,
    enable_memory=True
)

# Use for complex analysis
response = agent.run(
    input="Analyze this data and provide insights...",
    user_id="analyst123",
    session_id="analysis1"
)

Example 5: Custom Instructions¶

from bast_ai_agno.bp.agent_templates import create_rag_agent_template

# Create RAG agent with custom instructions
agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="medical_assistant",
    knowledge_base_name="medical_kb",
    custom_instructions=[
        "You are a medical assistant specialized in medication information.",
        "Always cite sources from the knowledge base.",
        "Provide clear and concise answers.",
        "If information is not in the knowledge base, recommend consulting a healthcare professional."
    ]
)

Best Practices¶

1. Use Templates for Common Patterns¶

# ✅ Good: Use template for common pattern
agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="kb_agent",
    knowledge_base_name="my_kb"
)

# ❌ Avoid: Manual setup for common patterns
# (unless you need custom resource setup)

2. Customize Instructions When Needed¶

# ✅ Good: Customize instructions for domain-specific needs
agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="legal_assistant",
    knowledge_base_name="legal_kb",
    custom_instructions=[
        "You are a legal assistant.",
        "Always cite legal sources.",
        "Provide disclaimers about legal advice."
    ]
)

3. Choose Appropriate Template¶

# ✅ RAG: For knowledge base queries
agent = create_rag_agent_template(...)

# ✅ Tools: For external integrations
agent = create_tool_agent_template(...)

# ✅ Conversational: For multi-turn chats
agent = create_conversational_agent_template(...)

# ✅ Reasoning: For complex problems
agent = create_reasoning_agent_template(...)

4. Use Factory for Custom Setups¶

# ✅ Use factory when you need custom resource setup
from bast_ai_agno.bp.agno_agents import AgnoAgentFactory

factory = AgnoAgentFactory("myapp")
# Custom setup with existing resources
agent = factory.create_openai_chat_agent(...)

Comparison: Templates vs Factory vs Configurations¶

Templates (Recommended for Common Patterns)¶

from bast_ai_agno.bp.agent_templates import create_rag_agent_template

agent = create_rag_agent_template(
    cat_name="myapp",
    agent_name="my_agent",
    knowledge_base_name="my_kb"
)

Pros: - ✅ Minimal setup - ✅ All resources configured automatically - ✅ Optimized for specific use cases - ✅ Ready to use immediately

Cons: - ❌ Less flexibility - ❌ Can't reuse existing resources

Factory (For Custom Setups)¶

from bast_ai_agno.bp.agno_agents import AgnoAgentFactory
from bast_ai_agno.bp.kb_resources import AgnoKnowledgeResources
from bast_ai_agno.dto.agent_configs import knowledge_agent_config

factory = AgnoAgentFactory("myapp")
kb_resources = AgnoKnowledgeResources("myapp")
knowledge = kb_resources.get_knowledge("my_kb")

agent = factory.create_openai_chat_agent(
    name="my_agent",
    model=factory._model_resources.get_open_ai_chat("gpt-4o-mini"),
    db=factory._storage_resources.get_agent_storage("my_agent"),
    knowledge=knowledge,
    config_raw=knowledge_agent_config
)

Pros: - ✅ Full control over resources - ✅ Can reuse existing resources - ✅ Flexible configuration

Cons: - ❌ More setup code - ❌ Need to manage resources manually

Configurations (For Reusable Configs)¶

from bast_ai_agno.dto.agent_configs import knowledge_agent_config

# Just get the configuration dictionary
config = knowledge_agent_config

Pros: - ✅ Reusable configuration dictionaries - ✅ Can combine with utilities - ✅ Flexible

Cons: - ❌ Need to set up resources separately - ❌ More code for complete setup

Additional Resources¶

Configuration Guide - Complete configuration guide
Configuration Schema - Schema reference
Examples - More examples
Agno Agents Overview

Template Implementation Details¶

Templates are located in bast_ai_agno.bp.agent_templates and use:

AgnoAgentFactory: For creating agents
Resource Managers: For getting models, storage, knowledge, memory
Preset Utilities: For optimized configurations
Agent Configurations: For base configurations

All templates return fully configured Agent instances ready to use.