From gdd
Guide for when to apply the GDD workflow — read at session start to understand when GDD applies and what to do first
How this skill is triggered — by the user, by Claude, or both
Slash command
/gdd:using-gddThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
<SUBAGENT-STOP>
GDD enforces a diagrams-first development process: update architecture and flow diagrams before writing code.
| Task type | GDD required? |
|---|---|
| Implement a new feature | YES — invoke gdd:plan skill |
| Add a new API endpoint or route | YES — invoke gdd:plan skill |
| Refactor a module's structure | YES — invoke gdd:plan skill |
| Add a new component / service | YES — invoke gdd:plan skill |
| Fix a bug in existing code | NO |
| Fix a typo or rename | NO |
| Write or update tests | NO |
| Answer a question | NO |
| Update documentation | NO |
| Update a config value | NO |
Rule: If the task adds new behavior or changes how modules interact — GDD applies. If it corrects something that was supposed to work already — GDD does not apply.
Is this a feature/new-functionality task?
Does docs/gdd/ exist with at least one flow-*.md AND one arch-*.md?
gdd:init first, then invoke gdd:plangdd:plan skill directly with the user's requirement as the argumentBriefly inform the user, then ask for confirmation before invoking gdd:init (since it scans the project and creates files):
This project doesn't have GDD diagrams yet. I'll need to map the current architecture first via
gdd:init, then rungdd:planto design the changes.Shall I start with
gdd:initnow?
If the user confirms, invoke the gdd:init skill. After it completes, invoke the gdd:plan skill with the original requirement as the argument.
Do not wait for user confirmation. Immediately:
gdd:plan for your requirement."gdd:plan skill with the user's requirement as the argument.If the user says "just implement it, skip the diagrams" — respect that.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub vivaxy/claude-code-plugins-vivaxy