Skip to content
Get packs

Agentic Engineering Workflow

Guides an agent to plan work in small verified steps, evaluate before and after changes, and route tasks to cost-appropriate models.

Source Download skill
Repository
affaan-m/ECC
Stars
295,127
Licence
MIT
Files
1 file

What it does

This is a set of working rules, written in Japanese, that tells an AI coding agent how to run a software project. Before changing code, the agent defines what 'done' means and runs a baseline test, then splits the job into small pieces that can each be checked on their own. It also suggests using cheaper AI models for simple edits and stronger ones for architecture and root-cause analysis, and it asks the agent to track cost, retries and time per task. There are no scripts or external tools; it only changes how the agent behaves.

Great for

  • Developers who let AI agents write most of their code and want stronger review and measurement
  • Teams wanting a checklist for controlling AI-generated changes and their cost
  • Japanese-speaking engineers who want the guidance in their own language

Skip it if

  • Beginners looking for a ready-made app or tool: it is a workflow guide, not a program, and the evaluation steps still need a real test setup from you
  • Projects with no tests or evaluations: the 'evaluation first' loop cannot run without them
  • Anyone expecting the skill to choose or switch models automatically: it only describes a routing policy the agent must follow

Try asking

  • “Plan this feature as small verifiable steps with clear completion criteria before you start coding”
  • “Set up a capability and regression evaluation for this bug fix, then run the baseline first”
  • “Route these tasks: classify the issues with a small model, refactor the module with a mid-tier model, design the architecture with the largest model”
  • “Review this AI-generated pull request focusing on invariants, error boundaries and security assumptions rather than style”
Step-by-step guide: how to use Agentic Engineering WorkflowInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    agentic-engineering.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agentic-engineering.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agentic-engineering.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Provide your existing test or evaluation commands so the agent can run the baseline loop instead of just describing it
  • Ask the agent to report the cost table (model, token estimate, retries, time, outcome) at the end of each task
  • The skill is written in Japanese; ask for English output explicitly if your team works in English

What's inside

  • SKILL.md

Redistributed unmodified from affaan-m/ECC under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.