Claude Code SWE #3: Skills and Workflow engineering

Happy New Year! Welcome to the third Claude Code meetup. Following our sessions on setup and project structure, we are diving deep into the engine that powers effective agents: Skills and Workflows.

**************************************
OBS - We are moving to Luma, register here
https://luma.com/k1o8dtpx
**************************************

## What We'll Cover

As agentic coding matures, we are seeing a shift in focus: from Prompting to Context, and now to Workflow Engineering. This session is about moving beyond simple instructions and building high-performance agentic systems that are modular, verifiable, and scalable.

## Skills: The Atomic Units

We’ll start by defining the "Skill." Whether you are following the official Anthropic standard or community-driven structures, a skill is a self-contained capability. We will explore:

• The Skill Definition: How to structure skills so an agent knows exactly when and how to use them.
• **SKILL.md:** We’ll introduce the concept of using documentation as a boundary. A skill should have a verifiable goal—did the skill actually execute what it was designed to do?

## Workflows: The Orchestration Layer

Workflows are often the "missing link" in agentic setups. If a skill is a tool, the workflow is the blueprint. We will discuss:

• **WORKFLOW.md:** How to move logic out of the system prompt and into dedicated workflow files.
• Logic Frameworks: How to orchestrate skills, slash-commands, and tools into a cohesive sequence.
• Beyond the Basics: Moving from simple prompt chaining to sophisticated feedback loops and routing.

## Separation of Concerns: Skill vs. Workflow

One of the biggest challenges in agentic coding is knowing where to define behavior. We will demonstrate why separating these two layers is critical:

• Skill Validation: Did the skill achieve its specific goal?
• Workflow Validation: Did the agent use the correct components (skills, tools, context) in the right order to solve the larger problem?

## Why This Matters

In the beginning, a long system prompt feels like enough. But as your agent grows in complexity, that prompt becomes a "black box" that is hard to debug and impossible to scale. By treating Workflows and Skills as distinct engineering artifacts, you gain the ability to test, iterate, and swap components without breaking the entire system. This session provides the patterns needed to build agents that don't just "guess," but follow a reliable process.

## Who Should Come

• Developers looking to move beyond basic Claude Code usage.
• AI Engineers interested in the transition from prompting to workflow engineering.
• Architects building complex agentic systems who need better mental models for modularity.
• Anyone curious about the latest Anthropic best practices for agentic loops.

No preparation is needed. We will review the official Anthropic documentation on Agent Skills and Workflows, then look at practical, real-world implementations using the `.md` framework.

## Beyond Claude Code

While our examples use Claude Code, the architectural principles of separating "capabilities" (Skills) from "logic" (Workflows) are universal. These patterns are directly applicable to the Gemini API, OpenAI Swarm, or any custom agentic framework you are building today.