visualstorming
Skill2 starsChanged 6 months agoMITView on GitHub
Adds an optional HTML-first visual reference for brainstorming with mockups, diagrams, wireframes, and side-by-side visual choices. Use when a discussion would be easier to resolve visually, especially for UI layouts, flows, architecture diagrams, spatial relationships, visual hierarchy, or design direction comparisons.
In this repository17 other files it gives its agents.
- create-nextjs-specificationSkill
- create-product-demoSkill
- find-github-repoSkill
- find-npm-packageSkill
- image-styleSkill
- nextjs-route-scaffoldSkill
- perfect-promptSkill
- place-meSkill
- review-meSkill
- setup-aeo-geoSkill
- setup-harnessSkill
- songwritingSkill
- step-by-stepSkill
- stock-analysisSkill
- study-meSkill
- teach-meSkill
- three-linesSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
21 rules. About 0 are common, 19 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈19
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use local HTML as a visual reference during brainstorming.
- Distinctive: line 10
Do not invoke writing-plans, create implementation plans, or start coding because this skill was used.
- Distinctive: line 14
When upcoming work likely benefits from visual treatment, offer visualstorming once before creating browser artifacts.
- Distinctive: line 20
Wait for the user response.
- Distinctive: line 24
Use browser visuals when the content itself is visual:
- Distinctive: line 32
Use text when the answer is mainly words:
- Distinctive: line 43
If not already approved, ask the consent gate question and stop.
- Distinctive: line 44
Create a local visual artifact in the current project, usually HTML.
- Distinctive: line 46
Ask one focused visual question.
- Distinctive: line 53
Prefer small HTML fragments or simple self-contained HTML files.
- Very specific: line 54
Use stable filenames that describe the screen, such as layout-options.html.
- Distinctive: line 55
Do not reuse filenames for materially different screens.
- Distinctive: line 56
Keep each screen focused on one decision.
- Distinctive: line 59
Avoid decorative complexity unless visual style is the decision being tested.
- Distinctive: line 86
Avoid broad questions like "Thoughts?", "What do you think?", or "Is this good?"
- Distinctive: line 90
Use the best browser mechanism available:
- Distinctive: line 92
If an in-app browser tool exists, open the local HTML file or localhost URL there.
- Distinctive: line 93
If a visual reference server exists, write the screen to its content directory and read its event log on the next turn.
- Distinctive: line 94
If no browser tool is available, create the HTML artifact and provide the local path or URL.
- Distinctive: line 100
Stop using visualstorming when the user declines, remaining decisions are conceptual, the visual choice has enough signal, or browser setup …
- Very specific: line 102
When the discussion is complete and you transition to implementation planning, remove visualstorming artifacts that were created for that …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Visualstorming
Use local HTML as a visual reference during brainstorming. This is a tool, not a mode: use it only for questions the user can answer better by seeing options than by reading text.
Do not invoke writing-plans, create implementation plans, or start coding because this skill was used. Visualstorming ends when the visual question is answered or the user declines visual help.
Consent Gate
When upcoming work likely benefits from visual treatment, offer visualstorming once before creating browser artifacts.
The offer must be its own message, with no context summary or clarifying question mixed in:
Some of this may be easier to discuss visually. I can open a local HTML view with mockups, diagrams, comparisons, or layout options as we go. It can use extra tokens and may require a local URL. Want to use visualstorming?
Wait for the user response. If they accept, use browser visuals only when useful. If they decline, continue text-only. If they already asked for visualstorming, browser mockups, visual options, wireframes, diagrams, or a visual comparison, treat that as consent.
Browser Or Text
Use browser visuals when the content itself is visual:
- UI mockups, wireframes, navigation structures, component layouts
- Side-by-side layout, color, typography, density, or design direction comparisons
- Architecture, data-flow, relationship, state-machine, or process diagrams
- Spatial questions where arrangement, grouping, or hierarchy matters
- Design polish questions about spacing, emphasis, rhythm, or look and feel
Use text when the answer is mainly words:
- Requirements, scope, constraints, success criteria
- Conceptual tradeoffs, pros/cons, API choices, data model choices
- Naming, copy, terminology, acceptance criteria
A UI topic is not automatically visual. "What should this wizard do?" is text. "Which wizard layout works better?" is visual.
Workflow
- Decide if the current question is visual.
- If not already approved, ask the consent gate question and stop.
- Create a local visual artifact in the current project, usually HTML.
- Open it in a local browser using the environment's browser capability.
- Ask one focused visual question.
- On the next user turn, combine their text response with any browser interaction data.
- Continue with another visual artifact only if the next question benefits from seeing it.
Artifact Rules
- Store all generated design artifacts under
.visualstorming/. - Prefer small HTML fragments or simple self-contained HTML files.
- Use stable filenames that describe the screen, such as
layout-options.html. - Do not reuse filenames for materially different screens; preserve previous options.
- Keep each screen focused on one decision.
- Label choices clearly with stable IDs such as
layout-aorflow-minimal. - Make visuals legible at desktop and mobile widths.
- Avoid decorative complexity unless visual style is the decision being tested.
Interaction Pattern
For choice-based screens, make options selectable and record the selection when the environment supports it. If structured browser events are unavailable, ask the user to reply in chat with the option label.
Recommended option markup:
<section class="options">
<button data-choice="layout-a" aria-label="Choose layout A">
<strong>Layout A</strong>
<span>Compact sidebar with dense controls</span>
</button>
<button data-choice="layout-b" aria-label="Choose layout B">
<strong>Layout B</strong>
<span>Top navigation with wider content area</span>
</button>
</section>Ask focused questions:
- "Which layout feels easier to scan?"
- "Which diagram matches your mental model?"
- "Which visual direction should we refine?"
Avoid broad questions like "Thoughts?", "What do you think?", or "Is this good?"
Browser Session Notes
Use the best browser mechanism available:
- If an in-app browser tool exists, open the local HTML file or localhost URL there.
- If a visual reference server exists, write the screen to its content directory and read its event log on the next turn.
- If no browser tool is available, create the HTML artifact and provide the local path or URL.
For server-backed references: save screen_dir, state_dir, URL, and session directory; confirm the server is alive before each new screen; read interaction events only after the user responds in chat. Chat text is primary feedback; browser events are supporting data.
Stop Conditions
Stop using visualstorming when the user declines, remaining decisions are conceptual, the visual choice has enough signal, or browser setup costs more than the decision warrants.
When the discussion is complete and you transition to implementation planning, remove visualstorming artifacts that were created for that discussion from .visualstorming/.
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
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.

