agentic-engineering

Coordinate AI-driven engineering with eval-first execution and tiered model routing.

Updated Apr 4, 2026

Search Tags:#automation #decomposition #risk-management #agentic #model-routing #engineering-workflow #eval-first

Ready to run

npx skills add https://github.com/mitul-bhatia/Vibes --skill agentic-engineering-mitul-bhatia

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agentic-engineering
Source: https://github.com/mitul-bhatia/Vibes/tree/main/.github/skills/agentic-engineering
Command: npx skills add https://github.com/mitul-bhatia/Vibes --skill agentic-engineering-mitul-bhatia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Operates as an agentic engineer performing implementation work with governance and risk controls to maintain quality while accelerating delivery.

Core Features & Use Cases

Eval-first execution: plan, run, and evaluate implementations with regression checks.
Task decomposition: break work into independently verifiable units with a single dominant risk per unit.
Model routing: route tasks to Haiku, Sonnet, or Opus based on complexity.
Session strategy: maintain session continuity across related units and refresh after milestones.
Review focus for AI-generated code: emphasize invariants, error handling, security assumptions, and risk mitigation.

Quick Start

Decompose a feature into agent-sized tasks, route work by complexity, and run eval-first checks after each unit.

Frequently Asked Questions about agentic-engineering

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

FAQPage Schema

How do I decompose complex software features into agent-sized tasks for AI engineering?▼

Eval-first execution plans, runs, and evaluates AI implementations with regression checks after each unit. This workflow maintains quality during rapid feature delivery by enforcing baseline evaluation checks before completing any independently verifiable task.

What is cost-aware model routing for AI engineering workflows?▼

Agentic engineering governs AI-generated code by emphasizing invariants, error handling, security assumptions, and risk mitigation. It applies structured governance and risk controls across multi-step tasks to maintain quality while accelerating delivery.

How do I manage session continuity across multi-step AI engineering tasks?▼

Agentic engineering suits complex software projects requiring rapid feature implementation, rigorous verification, and structured governance across multi-step tasks. It enforces independently verifiable units, tiered model routing, baseline checks, and clear done conditions with risk controls.

What's the best way to verify AI-generated code in agentic engineering workflows?▼

Decompose a feature into agent-sized tasks, route work by complexity to Haiku, Sonnet, or Opus, and run eval-first checks after each unit. This coordinates AI-driven engineering work while maintaining session continuity and enforcing risk controls.