brainstorming-gate

Brainstorming Gate

Overview

Turn ideas into designs through collaborative dialogue before writing code.

Core principle: No implementation without an approved design. Every project, regardless of perceived simplicity.

Do NOT invoke any implementation skill, write any code, scaffold any project, or take any implementation action until you have presented a design and the user has approved it. This applies to EVERY project regardless of perceived simplicity.

Anti-Pattern: "This Is Too Simple To Need A Design"

Every project goes through this process. A utility function, a config change, a single-file script — all of them. "Simple" projects are where unexamined assumptions cause the most wasted work. The design can be short (a few sentences for truly simple projects), but you MUST present it and get approval.

Process Flow

digraph brainstorming {
    "Explore project context" [shape=box];
    "Ask clarifying questions" [shape=box];
    "Propose 2-3 approaches" [shape=box];
    "Present design" [shape=box];
    "User approves?" [shape=diamond];
    "Save design doc" [shape=box];
    "Invoke strategic-planning" [shape=doublecircle];

    "Explore project context" -> "Ask clarifying questions";
    "Ask clarifying questions" -> "Propose 2-3 approaches";
    "Propose 2-3 approaches" -> "Present design";
    "Present design" -> "User approves?";
    "User approves?" -> "Present design" [label="no, revise"];
    "User approves?" -> "Save design doc" [label="yes"];
    "Save design doc" -> "Invoke strategic-planning";
}

Terminal state: invoke strategic-planning. Do NOT invoke development, architecture, or any other implementation skill. The ONLY next step is strategic-planning (to create the implementation plan).

Checklist

1. Prior art search — search T3 for prior decisions on this topic before exploring files or asking questions: mcp__plugin_nx_nexus__search(query="{feature or topic}", corpus="knowledge", limit=5 If a prior decision exists, surface it immediately — either re-use it (if still valid) or acknowledge it explicitly before proposing alternatives. Don't re-litigate settled decisions without knowing they were settled. If no prior decision exists, proceed to file and commit exploration.
2. Explore project context — check files, docs, recent commits
3. Ask clarifying questions — one at a time, understand purpose/constraints/success criteria
4. Propose 2-3 approaches — with trade-offs and your recommendation
5. Present design — scaled to complexity, get user approval after each section
6. RDR status check — Scan the user request, design doc, and relay task for the pattern RDR-\d+. For each match:
   • mcp__plugin_nx_nexus__memory_get(project="{repo}_rdr", title="NNN"
   • If status is not accepted or closed, warn the user: "RDR-NNN is still {status}. Run /nx:rdr-gate NNN and /nx:rdr-accept NNN before planning implementation."
   • If the lookup fails or returns no result, warn and proceed (fail-open).
   • If no RDR-\d+ pattern is found, proceed normally.
7. Write design doc — save to T2 memory via memory_put(project="{repo}", title="design-{topic}.md", content="...")
8. Transition — invoke strategic-planning skill to create implementation plan

Key Principles

• One question at a time — do not overwhelm with multiple questions
• YAGNI ruthlessly — remove unnecessary features from all designs
• Explore alternatives — always propose 2-3 approaches before settling
• Incremental validation — present design sections, get approval before moving on

Red Flags — STOP and Restart

• About to write code without a design
• "This is too simple for a design"
• "I already know what to build"
• Skipping questions because the answer seems obvious
• Jumping to implementation after one question

All of these mean: Stop. Follow the process.

Relay Template (Use This Format)

When invoking the strategic-planning skill after design approval, use this structure:

## Relay: strategic-planner

**Task**: Create phased implementation plan from approved design.
**Bead**: [ID] (status: [status]) or 'none'

### Input Artifacts
- nx store: [prior art or related decisions from T3, or "none"]
- nx memory: [session state, e.g. "{repo}/design-{topic}.md", or "none"]
- Files: none (design is in T2 memory)

### Deliverable
Phased implementation plan with beads for task tracking

### Quality Criteria
- [ ] Every task has exact file paths and test commands
- [ ] TDD: failing test before implementation in each task
- [ ] Dependencies between tasks are explicit

REQUIRED SUB-SKILL: Use nx:strategic-planning after design approval to create the implementation plan.