Brainstorming & Planning

•
12
•

Facilitates structured problem-solving by generating diverse technical options and refining them into actionable implementation plans.

The Brainstorm skill enhances Claude's ability to navigate ambiguous build or fix requests by applying a structured 'Diverge → Converge' pattern. It helps users identify hidden constraints, explore a wide breadth of technical approaches—ranging from quick wins to bold architectural shifts—and then prune those options into a focused, timeboxed roadmap. Whether you are stuck on a complex bug or starting a new feature without a clear stack, this skill ensures you consider all trade-offs and risks before committing to a single development path.

主要功能

01Automated constraint clarification for time, stack, and risk management
02Brief expectation alignment for optimized code and PR reviews
0312 GitHub stars
04Rapid generation of 5-10 varied technical approaches for any task
05Structured Diverge/Converge methodology for better decision-making
06Trade-off analysis comparing impact versus effort for chosen candidates

使用场景

01Evaluating multiple architectural approaches before beginning a build
02Clarifying ambiguous feature requests with missing technical constraints
03Unblocking development when stuck on a single, non-optimal solution
name description
brainstorm
Use when generating options or clarifying a build/fix request—pull constraints from the user, list varied approaches, then narrow with them to pick a plan. For code reviews, use briefly to confirm expectations, then hand off to the agent flow.

Brainstorm

Overview

Generate many concise options first, then converge with the user. Favor breadth, then prune to what fits scope, time, and constraints.

When to Use

  • Any build/fix request that lacks explicit constraints (scope, stack, perf, deadlines).
  • User asks for ideas, approaches, pros/cons, or “what are my options?”
  • Early in a task to choose direction or unblock.
  • When stuck or tunneling on one solution.
  • Before a code/PR review: confirm expectations (depth, checks to run) then proceed with the agent default review flow.

Core Pattern (Diverge → Converge)

  1. Clarify must-haves/constraints in 1–3 bullets (time, stack, budget, risk).
  2. Diverge: list 5–10 varied options (mix quick wins, bold bets, hybrids). One line each.
  3. Cluster and trim: group similar items, drop weak/out-of-scope ones.
  4. Select 1–3 candidates with rationale (impact vs effort vs risk).
  5. Ask for a pick or permission to detail the top choice.
  6. If chosen, expand into a concrete next-step plan (3–6 steps, timeboxed).

Tips

  • Keep options short; avoid over-detailing during divergence.
  • Include at least one low-risk/fast path and one higher-upside path.
  • Call out key trade-offs (time, complexity, reliability).
  • If info is missing, state assumptions and invite corrections.
  • Avoid anchoring: shuffle or avoid numbering during divergence; number only when converging.

Quick Reference

  • Start: “What constraints or must-haves should I respect? (time/budget/stack/risk)”
  • Diverge: bullet 5–10 options, 1 line each.
  • Converge: pick 1–3 best with why; ask which to pursue.
  • Expand: provide next steps for the chosen option.

Red Flags

  • Only one idea offered (no divergence).
  • Options are near-identical or all high-risk.
  • Deep detail before the user chooses a direction.
  • Ignoring stated constraints or assumptions.