jnemargut/flowchart-kit/flowchart

Claude Code and agent skill by jnemargut/flowchart-kit

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Build low-fi, marker-style flowcharts, sticky-note boards and an infinite canvas (frames, steps, decisions, stickies, stamps, links, rough charts, and cards showing storyboards or wireframes) as a flowchart.json file, validate it, and open the canvas editor. Use when the user asks for a flowchart, a user flow, a journey or process map, a sticky-note board, a brainstorm, branching paths of storyboards or wireframes, or one board to think through or present a product idea.

Install

npx vskill@latest install jnemargut/flowchart-kit/flowchart

SKILL.md

Flowcharts and a canvas, built by your agent

You turn a designer's loose thinking ("map what happens when an order runs late, and put the storyboard panel next to where trust breaks") into a *.flowchart.json board: frames holding flows of steps and decisions, stickies, stamps, links, and cards that show storyboards or wireframes. The tool lays the flows out for you, validates the file, and opens a canvas where the designer drags things around, adds more, and presents it. Everything is drawn in a hand-drawn marker style, so it reads as thinking, not as a final design.

Running the tool

Everything runs through one bundled script in this skill's folder (it needs Node.js 18+ and nothing else):

bash
node "${CLAUDE_SKILL_DIR}/scripts/flowchart.mjs" <command>

${CLAUDE_SKILL_DIR} is the folder this SKILL.md is in. If your agent doesn't fill it in, use that folder's path. Below, `fc` is short for that whole command.

Workflow

  1. Know the vocabulary. Don't guess. fc vocab lists node types, sticky colors, stamps, and frame and

link properties. Full list: references/vocabulary.md.

  1. Look for cards nearby. fc cards lists the storyboards and wireframes in the folder with their panel

and screen ids, so you can put the real thing on the board instead of describing it.

  1. Write the file: <name>.flowchart.json (fc new <file> makes a starter). Shape:

references/format.md. Worked example: references/example.md.

  1. Validate and fix until clean: fc validate <file>. Errors say exactly what to change ("did you mean …").

Warnings catch real problems (a decision with one way out, unlabeled answers, steps that should be stickies).

  1. Check it reads on its own: fc critique <file>. It tells you when a board is a pile of thoughts

instead of an argument: no answer up front, frames that name a topic, stickies of equal weight in a heap. Fix what it finds (a brainstorm wall can ignore it).

  1. Look at what you made. fc render <file> writes <name>.png (<file>#<frame id> renders one

frame, for a closer look). Read it and fix anything crowded or confusing before you say you're done.

  1. Open the editor in the background (it keeps running): fc dev <file>. Tell the designer the URL.

Their tweaks save into the same file in real time.

  1. Iterate on the same file. Re-read it before each edit, because the designer may have changed things.

Their touches are theirs: layout (nudges and sizes), canvas (where frames sit), shapes (their drawings, on the board or in a frame) and markup (Play-mode sharpie). Keep them unless asked. You can read them, though: a red circle or a scribbled word is often feedback for you.

  1. Present: frames are slides, in present order. fc export <file> --pdf (board plus a page per frame)

or --pptx (a slide per frame, with each frame's notes as speaker notes).

Write it plainly

Everything you write here (frame titles and descriptions, steps, stickies, labels on arrows, speaker notes) should sound like you telling a teammate across the table. If it sounds like a headline, a slogan or a slide from a consultant, write it again.

  • Whole sentences, ordinary words. Say who does what. "People don't mind when a tool comes back a day late."

Not "Owners shrug at late."

  • One idea in a sentence. No semicolons. No colon bolting two ideas together. If there are two things to say,

write two short sentences.

  • Keep the small words (the, a, their, when, because, so). Leaving them out is what makes writing read like a

telegram.

  • Say the thing, don't name it. "We don't know how often tools come back late" is clearer than "The claim has

no number behind it."

  • The plainest word wins. "Use", not "leverage". "Show", not "surface". "Problem", not "pain point". "The

hard part", not "friction". "Depends on", not "rests on" or "hinges on". "Because", not "due to".

  • No clever turns. No metaphors, no slogans, no lists of three for rhythm. If a phrase is showing off, cut

it.

  • Numbers and names as they are. "8 of 12 people". "The Pay button". Not "most users" or "the primary action".
  • Short. A title is one plain sentence you could say out loud, about ten words at most. A description is one

or two short sentences. A sticky is a dozen words. A step in a flow is the one place for a clipped label: a few words, verb first ("Asks the barista").

  • Read it out loud. If you wouldn't say it that way to a friend, change it until you would.
Instead ofWrite
Owners shrug at late; hidden damage is what stops them lendingPeople don't mind late returns. They stop lending when tools come back broken.
The ask rests on one claim with no number behind itThe PM says people return things late. We don't know how often.
The fee acts after the loan, and never looks at the toolA late fee doesn't check whether the tool is broken.
Try first: check the tool at handover, let the app do the askingA cheaper idea to try first
Late, nobody tells them, so they ask a humanWhen the order is late, people ask the barista
Trust lostAfter this, he stops ordering ahead
Surface the real queue to reduce frictionShow people the real queue so they know how long it'll be

The same goes for the message you send when you hand it over.

Craft: what makes a board useful

A board should make sense to someone who wasn't there. They open it, they know what it's saying in ten seconds, and the rest backs that up. A pile of notes doesn't do that. So:

  • Say the answer first. The first frame's description says what you found: "Probably not. People don't

mind late returns. They stop lending when tools come back broken." If the board's title is a question, this is where it gets answered.

  • Every frame has a title that says something, and a description. description is a plain sentence or two

under the title: what this frame shows. The title is a short sentence too ("People don't mind late returns"), not a topic ("Research"). Someone reading only the titles should get the whole story.

  • Give it a reading order. List the frames in present in the order you'd tell it. They get numbered on the

canvas. Place them in rows of two or three with near, like a comic, so the board fits on a screen: a row left to right, then the next row "near": ["below", "<the first frame>"].

  • Put notes in groups, under headings. Loose notes are laid out in the order you write them. A short text

followed by stickies becomes a column with the text as its heading. So write a heading, its notes, the next heading, its notes. Two to four columns a frame. Keep what you found, what you're guessing and what's still open in separate columns. Never leave six stickies in a pile.

  • Put a note about a step next to that step. near ties a sticky to its step with a dotted line. Without it

nobody can tell what the note is about.

  • End with what happens next: what you'd do, what someone has to decide, or the questions still open.
  • Less. A dozen words a sticky, 5 to 15 things a frame (not counting headings and stamps), five frames or

so. Cut anything that isn't needed.

  • Frames are the unit. Group a board into frames by story beat (how it should go, what really happens,

what to try, what's still open), each named for its point. Each becomes a slide. Place new ones by relation, never coordinates: "near": ["right of", "happy"], and the canvas finds the nearest free spot.

  • Never write coordinates. Flows lay themselves out (left to right, or "dir": "down"). You say what's

connected; the layout does the rest, and the designer's drags are kept on top.

  • Steps are short. A few words, verb first: "Checks the app", "Asks the barista". Anything longer is a

sticky beside it ("near": "<step id>").

  • Decisions have answers. A diamond gets a labeled link for each answer ("yes", "no", "after 10 min").
  • Show the unhappy path. The flow people actually take (waiting, asking a human, giving up) is where the

insight is. Give it its own frame.

  • Teal means the product. "product": true marks the steps where the product shows up; the rest is the

person's own world. Like in a storyboard, the contrast is the point.

  • Stickies carry the thinking. Pink for questions and worries, blue for ideas, green for what works, gray

for parked or assumed, yellow for everything else (what you found).

  • Stamps point at things. A cursor on a wireframe card shows the tap, a star marks the best idea, a

frown marks where it hurts: { "type": "stamp", "icon": "cursor", "near": "status", "at": [0.5, 0.8] }.

  • Numbers, roughly. When a number makes the point ("most people drop off at the cart", "8 of 12 asked

about the ETA"), put a chart beside it: { "type": "chart", "kind": "funnel", "text": "Where people drop off", "data": [["Browse", 1200], ["Cart", 640], ["Paid", 210]], "highlight": "Cart" }. Kinds: bar, hbar (long labels), line (over time), funnel, pie, donut (parts of a whole, a few slices). Do the math yourself first (with code if there's a file); the chart shows the two or three numbers that matter, not the whole sheet. highlight the one thing to look at and put the takeaway on a sticky next to it.

  • Link out. A ticket, a doc or a prototype is a link card ("url"), or a url on any step or frame.
  • Show, don't describe. If a storyboard panel or a wireframe screen exists for a moment, put it on the board

as a card and link to it from the step ("style": "dashed" reads as "see this").

  • Speaker notes. A frame's notes say what to point out when it's on screen.
  • Emphasis is fine, sparingly. Any text can use **bold**, *italic*, __underline__ and

~~struck out~~ (designers get Cmd+B, Cmd+I and Cmd+U in the editor). Bold the one word that matters, strike what changed.

  • Looks are optional. Boxes take size (s, m, l, xl), fill, stroke and weight; arrows take

shape (curved, angled, straight), style (solid, dashed, dotted), head (end, start, both, none), color and weight. Use them to mean something (red for where it breaks), not to decorate.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
5

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/jnemargut/flowchart-kit/flowchart/badge)](https://verified-skill.com/skills/jnemargut/flowchart-kit/flowchart)

Something wrong with this skill or its page? Tell us.