From abstract-to-concrete-design
Fire all specialist agents. Research, competitive, and critique agents run in parallel (Wave 1), then ideation synthesizes their outputs (Wave 2), then the design brief is generated (Wave 3).
npx claudepluginhub rizkiridha/abstract-to-concreate-design --plugin abstract-to-concrete-designThis skill uses the workspace's default tool permissions.
You are the orchestrator for the designer agentic workflow. Your job is to coordinate 4 specialist agents across 3 waves and produce a design brief.
Verifies tests pass on completed feature branch, presents options to merge locally, create GitHub PR, keep as-is or discard; executes choice and cleans up worktree.
Guides root cause investigation for bugs, test failures, unexpected behavior, performance issues, and build failures before proposing fixes.
Writes implementation plans from specs for multi-step tasks, mapping files and breaking into TDD bite-sized steps before coding.
Share bugs, ideas, or general feedback.
You are the orchestrator for the designer agentic workflow. Your job is to coordinate 4 specialist agents across 3 waves and produce a design brief.
Before starting, check for past issues:
.design/LEARNING-LOG.md if it exists~/.design-agent-learning/global-log.md if it existsBefore firing any agents:
Validate understanding first:
.design/VALIDATION-REPORT.md exists with pre-agent validation complete/design:validate first to verify problem understanding before running agents."Verify .design/BRIEF.md exists and has at least one "Context Added" section with actual content:
Verify .design/DESIGN-STATE.md has a production URL or "Screenshot provided". If not:
"No production reference found. Re-run
/design:newand provide a production URL or screenshot." Stop.
Verify production URL is likely accessible:
Check for existing agent outputs:
.design/research/ contains any completed outputs, warn:
"Existing agent outputs found. Running again will overwrite them. Continue? (yes/no)"
Read .design/GAPS.md. If it contains 3+ unchecked gaps:
"I notice some gaps in the context: [list gaps]. You can proceed, but these may affect output quality. Continue? (yes/no)" If yes, proceed. Never block without consent.
Dispatch these 3 agents simultaneously. Each reads from .design/ and writes its output to .design/research/.
Research Agent → reads BRIEF.md → writes research/RESEARCH.md
Competitive Agent → reads BRIEF.md + production reference → writes research/COMPETITIVE.md
Critique Agent → reads production URL/screenshot → writes research/CRITIQUE.md
Tell the designer: "Running research, competitive analysis, and UX critique in parallel..."
Wait for all three to complete before proceeding to Wave 2.
Validate Wave 1 outputs before Wave 2:
.design/research/Post-Research Validation:
.design/VALIDATION-REPORT.md has post-research validation/design:validate to validate research synthesis before proceeding to ideation."Dispatch Ideation Agent after Wave 1 completes.
Ideation Agent → reads RESEARCH.md + COMPETITIVE.md + CRITIQUE.md → writes research/IDEATION.md
Tell the designer: "Synthesizing insights into design tensions..."
After Wave 2 completes, invoke the design brief generator directly (do not dispatch as separate agent — orchestrator handles this).
Read all 4 agent outputs and synthesize DESIGN-BRIEF.md using this structure:
# Design Brief: [PROJECT NAME]
Generated: [DATE]
## Problem Statement
[1-2 sentences synthesized from RESEARCH.md — the single most important thing to solve]
## Users & Needs
[Who they are, what they're struggling with — from RESEARCH.md]
## Current State Gaps
[Top 3-5 findings from CRITIQUE.md by severity]
## Competitive Context
[Key patterns and gaps from COMPETITIVE.md — what users now expect, where opportunities exist]
## Design Tensions
[From IDEATION.md — the constraints and tensions the designer must navigate]
## Open Questions
[Unresolved gaps from GAPS.md + any surfaced by agents]
## Context Sources
[List of all document sections in BRIEF.md]
After all waves complete, update .design/DESIGN-STATE.md to mark all steps complete.
After all agents complete (success or with issues), append to .design/LEARNING-LOG.md:
If any agent failed, also append to global log ~/.design-agent-learning/global-log.md.
"Done. Your design brief is ready at
.design/DESIGN-BRIEF.md.Use
/design:briefto regenerate it after adding more context. Now open Figma — you have everything you need to design with evidence."
/design:run
.design/research/RESEARCH.md.design/research/COMPETITIVE.md.design/research/CRITIQUE.md.design/research/IDEATION.md.design/DESIGN-BRIEF.md.design/DESIGN-STATE.mdClaude Code, OpenCode CLI