Building an AI agent isn’t just about reasoning anymore — it’s about action. Once your agent can think, the next challenge is enabling it to interact with the outside world through tools, structured workflows, and universal protocols.
In this session, we focus on AI agent tools MCP skills — the three essential layers that allow agents to operate, integrate, and scale in real-world environments. Using the deepagents library, we’ll show how to make these layers composable, reusable, and production-ready.
What You’ll Learn
In this hands-on session, we’ll extend a research agent using AI agent tools MCP skills with practical, real-world implementations.
Building on Session 3, you’ll learn how to define custom tools with structured schemas, connect your agent to external systems using MCP (Model Context Protocol), and package repeatable workflows into reusable skills.
You’ll also understand when to use tools vs MCP vs skills — and how they work together to create powerful, extensible AI systems, powered by technologies from SambaNova Systems.
What You’ll Build
During this session, we will enhance an existing AI agent using AI agent tools MCP skills to:
- Create and integrate custom tools with structured input/output schemas
- Connect to an MCP server to enable external data and system access
- Package a reusable workflow into a skill for repeated execution
- Combine tools, MCP, and skills into a single composable agent
- Build a production-style agent capable of real-world interaction
The final system will demonstrate how AI agent tools MCP skills work together — transforming a research agent into a fully capable, connected system.
What We’ll Cover
- Tools, MCP & Skills — The Landscape — Understand the role of AI agent tools MCP skills and how they complement each other
- Custom Tools Done Right — Learn schema design, structured I/O, prompt engineering, and error handling
- MCP: The Universal Protocol — Explore how MCP enables standardized communication between agents and external systems
- Skills: Packaging Workflows — Turn complex processes into reusable, modular skills using SKILL.md
- Putting It All Together — Extend your agent with AI agent tools MCP skills in a unified architecture
- Production Patterns — Best practices for building scalable, composable AI agent systems
Who Should Attend
This webinar is ideal for:
- AI engineers building real-world agent applications
- Developers working with LLM-powered systems
- Engineers exploring tool use, automation, and integrations
- Anyone interested in AI agent tools MCP skills and composable AI systems
Technologies & Concepts
- Deep Agents (
create_deep_agent()) - AI Agent Tools MCP Skills
- Model Context Protocol (MCP)
- Custom Tooling & Schema Design
- Skills & Workflow Packaging
- Composable AI Systems
- Python