agentleFS
Sign inStart free
drafaeldrafael/coding-harness/skills/brainstorm/SKILL.md

brainstorm

Skill1 starsChanged 6 months agoMITView on GitHub

Explore unresolved future behavior, architecture, ownership, features, or substantive alternatives through collaborative dialogue before implementation. Use when the user explicitly asks to brainstorm, explore options, or think through a future change, or when another active skill identifies consequential decisions that still need exploration. Use explain-code for neutral current-code walkthroughs, architect when explicitly invoked for implementation-ready structural design, and technical-writing when available to structure an agreed design.

We split this skill into the rules it gives an agent

12 rules. 0 are common, 9 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 155 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check project context first, then ask questions one at a time:

    line 15
  2. Distinctive:

    Gather context: - check files, docs, recent commits relevant to the idea

    line 17
  3. Distinctive:

    Ask questions one at a time: - prefer multiple choice when possible

    line 18
  4. Distinctive:

    Do not overwhelm with multiple questions.

    line 21
  5. Distinctive:

    After approach is selected:

    line 35
  6. Distinctive:

    Be ready to backtrack: if something doesn't make sense

    line 40
  7. Distinctive:

    Do not present entire design at once.

    line 42
  8. Very specific:

    Create an implementation plan only when the user requested one or approves it after the design.

    line 46
  9. Very specific:

    When architect is installed and the user explicitly invokes it, hand off the approved problem and constraints for caller-first types, …

    line 48
  10. Very specific:

    When technical-writing is installed and loadable, apply it to a substantial plan without reopening approved design decisions.

    line 50
  11. Distinctive:

    One question at a time: - do not overwhelm with multiple questions

    line 54
  12. Distinctive:

    YAGNI ruthlessly: - remove unnecessary features from all designs, keep scope minimal

    line 56

“Common” is measured: kept by at least 6 in 10 of the 155 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Brainstorm

Turn ideas into designs through collaborative dialogue before implementation.

Process

Phase 1: Understand the Idea

Check project context first, then ask questions one at a time:

  1. Gather context - check files, docs, recent commits relevant to the idea
  2. Ask questions one at a time - prefer multiple choice when possible
  3. Focus on: purpose, constraints, success criteria, integration points

Do not overwhelm with multiple questions. One question per message. If a topic needs more exploration, break it into multiple questions.

Phase 2: Explore Approaches

Once the problem is understood:

  1. Compare 2-3 approaches when several are credible and explain their trade-offs
  2. When one approach is clearly supported, explain why alternatives do not fit instead of inventing options
  3. Lead with the recommendation and present it conversationally, not as a formal document

When comparing approaches, put each option label on its own plain-text line in the form Option A: Name (recommended) or Option B: Name, followed by bullets for how it works and its trade-offs. Do not prefix option labels with Markdown heading markers, bullets, numbering, backticks, or emphasis markers, and do not wrap the response in a code fence. The labels must remain readable without Markdown rendering.

Phase 3: Present Design

After approach is selected:

  1. Break design into sections of 200-300 words each
  2. Ask after each section whether it looks right
  3. Cover relevant concerns - for example, architecture, components, data flow, failure handling, and validation
  4. Be ready to backtrack if something doesn't make sense

Do not present entire design at once. Incremental validation catches misunderstandings early.

Phase 4: Next steps

Create an implementation plan only when the user requested one or approves it after the design. When the host can write files and the user wants an artifact, use YYYY-MM-DD-<topic>.md; otherwise return the plan in the response.

When architect is installed and the user explicitly invokes it, hand off the approved problem and constraints for caller-first types, signatures, data structures, and module boundaries. Do not make this handoff mandatory or reopen approved choices.

When technical-writing is installed and loadable, apply it to a substantial plan without reopening approved design decisions.

Key Principles

  • One question at a time - do not overwhelm with multiple questions
  • Multiple choice preferred - easier to answer than open-ended when possible
  • YAGNI ruthlessly - remove unnecessary features from all designs, keep scope minimal
  • Explore real alternatives - compare credible options without inventing choices to fill a quota
  • Incremental validation - present design in sections, validate each
  • Be flexible - go back and clarify when something doesn't make sense
  • Lead with recommendation - have an opinion, explain why, but let user decide
  • Duplication vs abstraction - discuss the trade-off only when both options are credible; prefer the simplest option that meets current needs

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free