mit einem Klick
using-development-skills
Use when starting any conversation - establishes how the development-skills plugin works and how to invoke its components on each platform (Claude Code, Codex). Read first.
Menü
Use when starting any conversation - establishes how the development-skills plugin works and how to invoke its components on each platform (Claude Code, Codex). Read first.
Use when user wants to create tests, generate test coverage, audit test quality, find untested code, or improve weak assertions. Use when user says write tests, test coverage, missing tests, or untested code.
Java development. Use for Java, Spring Boot, Maven, Gradle, JPA, Hibernate.
Python development. Use for Python, FastAPI, Pydantic, asyncpg, pytest, pandas, SQLAlchemy.
Swift development. Use for Swift, SwiftUI, UIKit, Vapor, SPM, XCTest, Combine.
Use when user wants to update requirements-dev.in with latest PyPI versions while preserving version patterns
Use when user wants to update requirements.in with latest PyPI versions while preserving version patterns
| name | using-development-skills |
| description | Use when starting any conversation - establishes how the development-skills plugin works and how to invoke its components on each platform (Claude Code, Codex). Read first. |
If dispatched as a subagent for a specific task: skip this skill.
If any skill might apply (even 1% chance), invoke it. Skills are the disciplined entry points — the meta-rule (spirit beats letter) covers the rest.
14 principles (0-13) + 1 meta-rule (spirit beats letter). Canonical: shared/iron-rules.md. Do not duplicate.
Classify the task before reading its content in depth:
shared/workflow.md # Tier selection) → follow the 6-step inline flow. Default on uncertainty → FULL.docs/plans/NNNN__YYYY-MM-DD__implementation_plan__slug.md + HOW-level Q&A locks (data shapes · edge cases · error semantics · contract boundaries · test scope · rollback).docs/chronicles/NNNN__YYYY-MM-DD__topic.md capturing initial decisions.staff-reviewer subagent review.Rules during FULL:
AskUserQuestion for discrete options; plain text otherwise; Codex fallback in references/codex-tools.md).Rules during LIGHT: Tier is qualitative (ambiguity / logic impact / new-pattern) — not file count. A 30-file mechanical rename is LIGHT; a 1-file new-caching-strategy is FULL. All Iron Rules still apply. Mid-execution discovery breaks LIGHT criteria → escalate to FULL per shared/workflow.md # LIGHT (final paragraph).
Routing: Bug fixes → development-skills:debugging. Test work → development-skills:create-test.
SKILL bodies use Claude Code tool names as canonical. On Codex, translate via references/codex-tools.md (Task → spawn_agent, AskUserQuestion fallback, staff-reviewer dispatch recipe, hooks, marketplace files).
User instructions (CLAUDE.md, AGENTS.md, direct requests) > development-skills > default system prompt.