skills.rest
gganbukim1
gganbukim1COMMUNITY
Explore all skills by this author →
GitHub

agentic-engineering

Plan eval-first coding workflows with regression checks and model tier routing.

Updated May 4, 2026
Search Tags:#agentic engineering#regression testing#model routing#eval-first#code decomposition#cost discipline
One-click install
npx skills add https://github.com/gganbukim1/myskills --skill agentic-engineering-gganbukim1
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agentic-engineering
Source: https://github.com/gganbukim1/myskills/tree/main/agentic-engineering
Command: npx skills add https://github.com/gganbukim1/myskills --skill agentic-engineering-gganbukim1

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you deliver reliable, production-ready implementations by forcing an eval-first workflow, careful decomposition, and cost-aware routing between model tiers.

Core Features & Use Cases

  • Eval-first execution loop: Baseline, run targeted regressions, implement, then re-run evals to quantify deltas and catch regressions.
  • Risk-focused decomposition: Break work into independently verifiable units with clear done conditions and a single dominant risk each.
  • Model tier routing: Use smaller models for narrow edits and classification, reserve larger models for refactors and architecture, and escalate only when the reasoning gap is clear.
  • Quality and safety guardrails: Prioritize invariants, edge cases, error boundaries, and security/auth assumptions over style-only debates.

Quick Start

Use the agentic-engineering skill to plan an implementation by defining completion criteria, decomposing tasks into eval-backed units, routing each step to the right model tier, and verifying changes with regression checks.

Frequently Asked Questions about agentic-engineering

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

FAQPage Schema
How do I reduce regressions when implementing multi-file code changes with AI agents?▼

Regression testing in agentic engineering requires an eval-first execution loop: establish a baseline, run targeted tests, implement changes, then re-run evals to quantify deltas and catch regressions before deployment.

What is eval-first agentic engineering and how does it improve implementation reliability?▼

Eval-first agentic engineering forces a workflow loop of baselining, implementing, and re-running comparisons. This verifies deltas against completion criteria and catches regressions during refactoring, architecture planning, and root-cause analysis.

How do I route tasks between different model tiers to maintain cost discipline during refactoring?▼

Cost-aware model tier routing uses smaller models for narrow edits and classification, reserving larger models for refactors and architecture. Escalate tiers only when a clear reasoning gap exists, ensuring cost discipline.

What is the best way to decompose complex coding tasks for independent verification?▼

Risk-focused code decomposition breaks work into independently verifiable units with clear done conditions and a single dominant risk each, prioritizing invariants, edge cases, error boundaries, and security assumptions over style.

Can I use this agentic engineering workflow for architecture planning and root-cause analysis?▼

Yes, the eval-first agentic engineering workflow applies to architecture planning and root-cause analysis across multi-file changes, using explicit completion criteria and regression checks to verify structural improvements.

Related Skills

Explore More Software Engineering Skills→
CANTAGESTUDIOCANTAGESTUDIO
3

refactor-test-safety-net

Identify missing tests and establish a minimal safety net before refactoring.

Community
Advanced
tchawla827tchawla827
1

agentic-engineering

Organize AI-driven implementation tasks into 15-minute units with quality gates.

Community
Advanced
eventua11yeventua11y
1

writing-tests

Automate E2E and unit test setup and execution with Playwright and Vitest.

Official
Intermediate
Oruga420Oruga420

agentic-engineering

Coordinate AI agents with eval-first execution, task decomposition, and cost-aware model routing.

Community
Advanced