| name | implement |
| description | Use when implementing a feature with clear, well-understood requirements. Quick planning via architect agent, then parallel agent development, then browser testing. Faster than craft — use when requirements don't need deep exploration. |
Implement
Fast end-to-end implementation pipeline. Quick planning, then development, then testing.
Pipeline: Architect → Develop → Browser Test → Report
Input
The user input is: $ARGUMENTS
Input Type Detection
-
Prompt File Number (e.g., 01, 07):
- Read from
.claude/prompts/{number}*.md
-
Direct Prompt Text:
- Use directly as feature requirements
-
Empty Input:
- Ask the user to provide a prompt file number or direct requirements
Phase 1: Architect
- Invoke
craft-skills:architect with the requirements
- Relay any clarification questions to the user
- Review the plan for alignment, reuse, and best practices
- Wait for user approval (allow iterations if changes requested)
If the architect raises too many ambiguities, suggest switching to craft-skills:craft for deeper exploration.
Phase 2: Develop
After plan approval, invoke craft-skills:develop with the approved plan.
Phase 3: Browser Test
After a successful build, invoke craft-skills:browser-test.
Phase 4: Report
Summarize:
- Plan decisions from Phase 1
- Files created/modified from Phase 2
- Test results from Phase 3
- Any open items
When to Use implement vs craft
| implement | craft |
|---|
| Planning | Quick architect agent | Deep brainstorming + spec review |
| User involvement | Medium — plan review | High — design questions, approach selection |
| Best for | Clear requirements, well-understood scope | Complex domains, unclear requirements |