Agentic Code — Testing skill for Claude Code

Agentic Code

A repository-resident development workflow for coding agents that read `AGENTS.md`.

Agentic Code adds an `AGENTS.md` entry point and a set of task, workflow, and skill files to a repository. The agent uses them to keep a well-scoped change on a direct path, route coordinated work through design and planning, load guidance only when it becomes relevant, and verify observable behavior before completion.

It does not generate an application by itself. It gives a coding agent a repeatable way to analyze, design, implement, test, and review changes in the repository where the files are installed.

[](LICENSE) [](https://agents.md) [](package.json)

What It Provides

AGENTS.md                      Repository entry point and operating boundaries
.agents/tasks/                 Definitions for analysis, design, implementation, QA, and review
.agents/workflows/             Coordinated workflow for structurally Medium/Large work
.agents/skills/                Reusable implementation, testing, and documentation rules
.agents/context-maps/          Task-to-skill selection map

The route is based on decision burden rather than file count:

Scale	Structural condition	Route
Small	One coherent outcome follows existing patterns within one responsibility boundary	Direct task execution
Medium	One coherent outcome coordinates across a boundary or requires a durable design decision	Design and Work Plan before implementation
Large	Multiple independently valuable outcomes require separate design decisions	PRD, separate design decisions, and Work Plan

File paths and affected layers are still inspected, but they are evidence about the change surface rather than the scale formula.

Quick