agentic-engineering

Orchestrate agentic engineering workflows with eval-first execution and model routing.

1|Updated Apr 6, 2026

Search Tags:#workflow #optimization #evaluation #decomposition #engineering #agentic

Ready to run

npx skills add https://github.com/vrcms/everything-qwen-code --skill agentic-engineering-vrcms

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agentic-engineering
Source: https://github.com/vrcms/everything-qwen-code/tree/main/.qwen/skills/agentic-engineering
Command: npx skills add https://github.com/vrcms/everything-qwen-code --skill agentic-engineering-vrcms

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill addresses the challenge of managing complex, multi-step engineering tasks where AI agents perform the heavy lifting, ensuring that human oversight remains effective and risk is minimized.

Core Features & Use Cases

Eval-First Execution: Ensures every task is measured against baseline and regression evaluations before and after implementation.
Task Decomposition: Breaks down large engineering goals into 15-minute, independently verifiable units.
Model Routing: Optimizes cost and performance by matching specific task complexities to the appropriate model tier (Haiku, Sonnet, or Opus).

Quick Start

Invoke the agentic-engineering skill to decompose the current feature request into verifiable units and establish the initial evaluation criteria.

Frequently Asked Questions about agentic-engineering

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I manage complex agentic engineering workflows to minimize risk?▼

Agentic engineering workflows orchestrate AI agents by enforcing eval-first execution, breaking goals into 15-minute verifiable units, and matching task complexity to appropriate model tiers like Haiku, Sonnet, or Opus.

How do I decompose large software development goals into verifiable units?▼

Decomposing software development goals involves breaking large engineering tasks into 15-minute, independently verifiable units to satisfy strict requirements for verifiable completion criteria and regression testing.

What is eval-first execution in AI-driven code generation?▼

Eval-first execution in AI code generation ensures every task is measured against baseline and regression evaluations before and after implementation to maintain high-quality output and rigorous risk management.

Can I optimize model routing costs for complex software development lifecycles?▼

Optimizing model routing costs for software development lifecycles involves matching specific task complexities to the appropriate model tier, such as Haiku, Sonnet, or Opus, balancing performance and expenditure.

Does agentic engineering work for multi-step tasks requiring regression testing?▼

Agentic engineering works for multi-step tasks by satisfying explicit requirements for verifiable completion criteria and regression testing, ensuring high-quality code generation across complex software development lifecycles.

When should I not use task decomposition for AI agents?▼

Task decomposition for AI agents may not suit simple, single-step engineering goals that lack strict requirements for verifiable completion criteria, baseline evaluations, and rigorous risk management.