Skip to main content
Starters are curated, deployment-ready project templates that help teams build and deploy agentic AI applications quickly and consistently.
  • Base Template: A foundational template (starter-agent-template-agno) that provides the core structure, configuration, and tooling.
  • Starter Kits: Specialized templates derived from the base template (e.g., starter-agent, starter-multi-agent) that implement specific agent patterns.
Teams can provision new repositories using any of these templates via Self Service, effectively bootstrapping their application with a production-ready setup.

​
Why use Starters?

Purpose

Provide a fast, reliable way to bootstrap agentic AI applications.

Consistency

Common project layout, configuration, integration, and tooling across all applications.

Extensibility

Easily extend on top of the template as per the use case.

​
Base Template

The Agno Base Template is the foundation for all other starters. It implements the core patterns for agent design, API exposure, observability, and configuration. Use this template if you want a clean slate to build a custom agent architecture while retaining standard platform integrations.

​
Available Templates

Repositories created from these templates automatically run the Repository Provisioning Workflow on first push. See CI/CD Workflows for details.
Quick scaffold using the Agno template: 
git clone https://github.com/bb-ecos-agbs/starter-agent-template-agno my-agent
cd my-agent
uv sync
uv run python -m src.main

​
Project Structure

The base template follows a standardized folder structure: 
├── .github/                   # CI/CD workflows
├── promptfoo_config/          # Evaluation configurations
├── prompts/                   # Centralized prompt definitions
├── providers/                 # AI provider configurations
├── src/
│   ├── main.py                # Application entry point
│   ├── agents/                # All agent implementations
│   ├── api/                   # FastAPI routers, schemas
│   ├── config/                # Configuration management
│   ├── logger.py              # Logging setup
│   └── __init__.py
├── tests/                     # Test files
├── redteam.yaml               # Red teaming configuration
├── pyproject.toml             # Dependencies (UV)
├── Dockerfile                 # Container definition
└── README.md

​
Standard Tooling

ComponentTool
Dependency ManagementUV
Web FrameworkFastAPI
ObservabilityLangfuse (via bb-ai-sdk) or any OTEL endpoint (optional)
AI GatewayBackbase AI Gateway (via bb-ai-sdk)
Evaluation & RedTeamingPromptfoo

​
Built-in CI/CD

All starter kits and templates come pre-configured with standard CI/CD Workflows. These automated pipelines handle:
  • Linting & Formatting: Enforcing code quality standards on every Pull Request.
  • Testing: Running unit and integration tests to ensure reliability.
  • Docker Builds: Building and publishing optimized container images.
  • Deployment: Standardized deployment strategies for releases.
Learn more about how to manage your agent’s lifecycle in the CI/CD Workflows section.

​
Available Starter Kits

These are specialized templates pre-configured with specific agent patterns. Use them to jumpstart development for common use cases.

Starter Agent (Level 0)

Single Agent: Basic instruction following, reasoning, and tool use. Great for simple tasks.

Multi Agent (Level 1)

Agent Teams: Patterns for delegation, collaboration, and coordination between multiple agents.

MCP Agent (Level 2)

(Coming Soon) Integration: Agents pre-wired with Model Context Protocol (MCP) for accessing banking services.

Knowledge Agent (Level 3)

(Coming Soon) RAG: Agents with knowledge base integration for document-based Q&A.
How to use: Select any of these starters (e.g., starter-agent, starter-multi-agent) as the repository_template in Self Service to provision a new repo with this code.