Advanced Agent Workflows and Skills

Design reliable AI agent workflows: patterns, skills, subagents, hooks, context management, evals and safe automation.

Start course →

Syllabus

Workflow Patterns

  1. Workflow or Agent?
  2. Prompt Chaining and Routing
  3. Parallelization and Orchestrator-Workers
  4. Evaluator-Optimizer Loops

Skills (SKILL.md)

  1. Anatomy of a Skill
  2. Descriptions and Progressive Disclosure
  3. Skills with Scripts

Subagents and Delegation

  1. Why Use Subagents
  2. Defining a Subagent
  3. Parallel Agents and Git Worktrees

Context and Memory

  1. The Context Budget
  2. Project Memory with CLAUDE.md
  3. Compaction and Handoff Notes

Hooks and Automation

  1. What Hooks Are
  2. A Guardrail Hook
  3. Headless Mode and CI

Tools, MCP and Permissions

  1. Designing Good Tools
  2. MCP in Workflows
  3. Permissions and Least Privilege

Reliability, Evals and Cost

  1. Verification Loops
  2. Evals for Agents
  3. Failure Handling and Stop Conditions
  4. Cost and Latency

Putting It Together

  1. Case Study: Automated Pull Request Review
  2. Revision: Cheat Sheet and Self-Check