06 design plan
Skill andrefabricio/prodotypor/files/.claude/skills/06-design-plan
An AI-assisted product development pipeline for Claude Code. Go from a product idea to Figma screens in 8 structured steps, with specialized AI agents handling product management, design research, and visual construction.
npx -y skills add andrefabricio/prodotypor --skill 06-design-planAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Step 6 of 8 — Design plan creation. Produces a Figma construction plan with screen inventory, batch organization, frame tree specs, and token sync tracking. Sub-commands: [continue] resume, [status] check progress, [update] force status refresh.
SKILL.md
5.6 KB, as published. Nobody here has run it
/06-design-plan — Design Plan Creation
Create the Figma construction plan that the figma-designer agent will execute.
Sub-command Router
Parse $ARGUMENTS:
- Contains
[status]→ Jump to Status Report section - Contains
[update]→ Jump to Force Status Update section - Contains
[continue]or empty → Jump to Execute section - Other text → Treat as specific batch/screen to plan
Prerequisites
Hard dependencies:
- Read
docs/workflow-status.md— verify step 05 (Design Doc) status ==completed - Read
docs/workflow-status.md— verify step 03 (PRD) status ==completed - Verify
docs/Design/{{PROJECT_NAME}}-design-document.mdexists with color + typography sections - Verify
docs/prds/*.mdexists with user stories
If prerequisites fail:
- If step 05 missing: "Step 06 cannot start. Missing: design document. Run
/05-design-docfirst." - If step 03 missing: "Step 06 cannot start. Missing: PRDs with user stories. Run
/03-prdfirst." - Do NOT proceed.
Soft dependencies:
- Figma file already created (nice but not required — plan can be written before file exists)
.claude/knowledge/figma-mcp-reference.md(MCP API patterns for call budgeting)
Execute
Phase 1: Gather All Inputs
- Read
CLAUDE.md— project context - Read
.impeccable.md— brand personality, design principles - Read
docs/Design/{{PROJECT_NAME}}-design-document.md— all tokens, component specs - Read all PRDs in
docs/prds/— extract user stories for screen mapping - Read
.claude/knowledge/figma-mcp-reference.md— API patterns, call budget - Read
templates/design-plan-template.md— output structure - If continuing, read existing plan to determine what sections are complete
Phase 2: Delegate to Design Planner Agent
Spawn the design-planner agent:
Create the Figma construction plan for {{PROJECT_NAME}}.
Design document: [content — tokens, components, principles] PRD user stories: [list with story IDs] MCP reference: [call budget, API patterns] Template: [content of
templates/design-plan-template.md] Focus: [from $ARGUMENTS if specific batch/screen]Requirements:
- Map every PRD user story to at least one screen
- Organize screens into batches by user flow (4-7 per batch)
- Write consistency rules referencing design doc token names
- Write inline frame tree specs for at least batch 1
- Include token sync log structure
- Include component instance reference table (empty, populated during construction)
- Include execution checklists per batch
- Save to
docs/Design/figma-design-plan.md
Pipeline Context
Before delegating, read and summarize prior step outputs:
- Step 03 output: read PRD user story IDs and titles for screen mapping
- Step 05 output: read design doc token names and component names (not full specs)
- Include summaries in the agent prompt as "Pipeline Context"
Context Management
The design plan agent reads the full design document (potentially 50KB+). To prevent context rot:
- Pass only the token tables and component names from the design doc, not full prose
- Pass PRD user story IDs and titles, not full Given/When/Then criteria
- Include figma-mcp-reference.md's Call Budget section, not the full file
Phase 3: Validate Output
- Verify
docs/Design/figma-design-plan.mdexists - Check: screen inventory has all PRD stories mapped
- Check: at least 1 batch with frame tree specs
- Check: consistency rules section exists
- Check: token sync log section exists
- If validation passes → proceed to Phase 4
- If validation fails → proceed to Error Recovery
Error Recovery
If validation fails:
- Identify the specific failure (e.g., "3 PRD stories have no mapped screen")
- Feed the error + partial output back to the agent:
"The design plan did not meet requirements: [specific failure]. Current output: [screen count, batch count, missing mappings]. Please fix: [specific instruction]."
- Re-validate after the retry
- If retry also fails, mark step as
blocked (reason)indocs/workflow-status.md
Phase 4: Human Review Gate
- Update
docs/workflow-status.mdstep 06 status toneeds-review - Report plan summary (screen count, batch count, spec coverage) to the user
- Prompt: "Review the design plan. Run
/06-design-plan [continue]to approve, or provide feedback." - Only
[continue]afterneeds-reviewtransitions tocompleted
Phase 5: Report
Step 06 — Design Plan: [status]
Output: docs/Design/figma-design-plan.md
Total screens: [count] across [batch count] batches
Batch 1: [screen count] screens ([flow name]) — specs: [written/pending]
Batch 2: [screen count] screens ([flow name]) — specs: pending
...
Components referenced: [count]
Consistency rules: [count]
Next step: /07-figma
Status Report
- Read
docs/workflow-status.md— extract step 06 section - If plan exists, report: screen count, batch count, spec completion
- Report which batches have frame tree specs vs pending
Force Status Update
- Read
docs/Design/figma-design-plan.md - Count screens, batches, specs written
- Check token sync log freshness
- Update
docs/workflow-status.mdstep 06