ai-workflow-engineering

Installation
SKILL.md

AI Workflow Engineering Pattern

When to Use This Skill

Creating reliable AI workflows that manage LLM uncertainty through structured phases, explicit constraints, and human checkpoints. This is the universal pattern that emerged across PDD, HumanLayer, and Agent SOPs.

The 5-Phase Universal Structure

All reliable AI workflows follow this pattern:

  1. Intake/Investigation - Validate inputs before processing
  2. Decomposition/Planning - Break work into components
  3. Iterative Execution - Do work with checkpoints
  4. Validation/Review - Test quality with metrics
  5. Decision Point - Human chooses next action

Creating a New Workflow

Parameters to Gather

Related skills

More from m31uk3/ai-skills

Installs
3
GitHub Stars
5
First Seen
Mar 1, 2026