Agent skill

Canvas

by nimbalyst in nimbalyst/nimbalyst

Author Nimbalyst Project Canvas boards (.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges.

MITAuto-check passed

Install Canvas

skills CLI
$ npx skills add nimbalyst/nimbalyst --skill canvas -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install nimbalyst/nimbalyst canvas --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/nimbalyst/nimbalyst.git skills-src && mkdir -p .claude/skills && cp -r skills-src/packages/extensions/canvas/claude-plugin/skills/canvas .claude/skills/canvas && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
canvas
GitHub stars
1.9k
Token cost
~2.6k tokens
SKILL.md length
997 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
MIT

At a glance

Author Nimbalyst Project Canvas boards (.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges.

  • The user asks for a canvas
  • SKILL.md covers When to use a canvas, File format, Cards and Layout, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
  • A spatial layout of documents

What it does

Canvas is an agent skill from nimbalyst/nimbalyst. Author Nimbalyst Project Canvas boards (.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges. Use when the user asks for a canvas, a board, a spatial layout of documents or mockups, a flow of screens, or a workspace overview.

Its SKILL.md is about 2.6k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: Nimbalyst - The open-source visual workspace for Claude Code, Codex, and OpenCode. Run multiple coding agents in parallel, edit their work visually in markdown, mockups, and… The licence is MIT.

When your agent uses it

  • The user asks for a canvas
  • A spatial layout of documents
  • A flow of screens
  • A workspace overview

Example prompts

  • “/canvas”

What it can do on your machine

Read from SKILL.md and the folder at commit 6f14adf. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are json).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Canvas loads about 2.6k tokens when it runs. Until then it costs about 83 tokens; SKILL.md has 997 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~83
When it runs · the whole SKILL.md, loaded when a task matches
~2.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from nimbalyst/nimbalyst at commit 6f14adf, republished under its MIT licence (© nimbalyst). 997 words, ~2,585 tokens.

Download SKILL.mdSave it as .claude/skills/canvas/SKILL.md (or your agent's skills folder).
name
canvas
description
Author Nimbalyst Project Canvas boards (.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges. Use when the user asks for a canvas, a board, a spatial layout of documents or mockups, a flow of screens, or a workspace overview.

Project Canvas

A .canvas file is a board: an infinite canvas whose cards can be sticky notes, text, images, frames, or live editors mounted on real workspace files and shared documents. A mockup card renders the mockup. A spreadsheet card renders the grid. A mindmap card renders the mindmap.

The editor is built into Nimbalyst, so any workspace can open a .canvas file you write. Write the JSON directly with the Write tool; there is no MCP tool for boards.

When to use a canvas

  • Laying out several existing files spatially — screens of a flow, docs in a workstream, a review board
  • A workspace or project overview that mixes documents, mockups, data, and notes
  • Anything where the point is the arrangement of real artifacts, not a drawing

Use Excalidraw for freeform drawings and architecture sketches. Use MockupLM for a single UI screen. A canvas is the thing that holds several of those at once.

File format

JSON Canvas 1.0 superset. Every spec field keeps its spec meaning, Nimbalyst data lives under x-nimbalyst, and unknown keys at any level are preserved verbatim so a board written by another tool survives a round-trip.

json
{
  "nodes": [],
  "edges": [],
  "x-nimbalyst": {
    "version": 1,
    "meta": {
      "name": "Onboarding flow",
      "description": "Sign-up through first document",
      "viewport": { "x": 0, "y": 0, "zoom": 1 }
    }
  }
}

meta.viewport is the board's deliberate home view — where a reader lands on open. Each user's own scroll position is stored separately, so setting this is a choice about the board, not about you. Omit it and the board opens fit-to-content.

Nodes

Every node needs id (unique, non-empty), type, and x / y / width / height. Geometry is written as integers; the editor rounds on save either way.

typeRequired payloadNotes
texttextPlain text, not markdown
filefileWorkspace-relative path
linkurlAlso carries shared-document URIs
group—Optional label, background, backgroundStyle

color is optional on nodes and edges: the presets "1" red, "2" orange, "3" yellow, "4" green, "5" cyan, "6" purple, or any #rrggbb. Anything else is ignored.

Array order is z-order. The last node in nodes paints on top — put a frame before the cards that sit inside it. Do not invent a z-index or rank field; the editor derives ordering from array position and never writes one to the file.

Serialization is canonical, not preserved

On save the editor rewrites keys into a fixed order, sorts edges by id, and rounds geometry. Your key ordering and edge ordering will change the first time someone opens the board. That is expected — write it readably and let the editor normalize.

Cards

What a card draws as is decided by x-nimbalyst.reference, not by the spec type. The spec type is chosen so a plain JSON Canvas reader still shows something true. Always write both.

x-nimbalyst.label sets the card's label.

File card — mounts the real editor
json
{
  "id": "login",
  "type": "file",
  "x": 0,
  "y": 0,
  "width": 640,
  "height": 440,
  "file": "design/mockups/login.mockup.html",
  "x-nimbalyst": {
    "reference": { "kind": "file", "path": "design/mockups/login.mockup.html" },
    "label": "Login"
  }
}

The path is workspace-relative and must appear in both file and reference.path. Any file type with a registered editor works — .mockup.html, .mindmap, .excalidraw, .csv, .md, and so on. The card mounts that editor live; it is not a screenshot.

Point a file card at a file that exists. A card for a missing path renders as an unresolved placeholder.

Doc card — a shared document
json
{
  "id": "prd",
  "type": "link",
  "x": 700,
  "y": 0,
  "width": 640,
  "height": 440,
  "url": "nimbalyst://doc/{orgId}/{documentId}",
  "x-nimbalyst": {
    "reference": { "kind": "doc", "uri": "nimbalyst://doc/{orgId}/{documentId}" },
    "label": "Product brief"
  }
}

Only write a doc card when you have a real org and document id. Do not guess them — prefer a file card.

Native cards

reference.kind: "native" with a nativeKind, paired with the spec type below:

nativeKindSpec typePayloadDefault size
stickytexttext, conventionally "color": "5"240 × 180
texttexttext320 × 200
imagelinkurl360 × 260
groupgrouplabel640 × 440
json
{
  "id": "note-open-question",
  "type": "text",
  "x": -260,
  "y": 0,
  "width": 240,
  "height": 180,
  "color": "5",
  "text": "Do we verify email before or after the first document?",
  "x-nimbalyst": { "reference": { "kind": "native", "nativeKind": "sticky" } }
}

A node with no x-nimbalyst.reference still renders — it falls back to its spec type, which is what makes a board from another tool useful rather than a wall of placeholders. Writing the reference is what gets you a sticky note instead of a plain text card.

A node whose type is outside the spec renders as an "unsupported card" placeholder and is written back untouched. Never use that as a way to smuggle in data.

Show full SKILL.md (349 more words)Show less

Layout

  • Snap to the 20px grid. Every x, y, width, height should be a multiple of 20. The editor snaps on drag, so an off-grid board shifts the first time someone touches it.
  • Use the default sizes above unless you have a reason. Reference cards default to 640 × 440 — big enough that the editor inside is legible without a resize.
  • Leave 60–100px of gutter between cards. Cards have toolbars above, comment badges to the right, and presence chips below; touching cards make all three collide.
  • Lay flows left-to-right, branches top-to-bottom. Read order is the layout's job.
  • Frames are plain rectangles. There is no parentId and children keep absolute coordinates. A frame does capture the nodes geometrically inside it when dragged, so size it to genuinely contain them, and put it earlier in nodes so it paints behind.
  • Give the board a meta.name. It is what the board is called in listings.

Edges

json
{
  "id": "login-to-dashboard",
  "fromNode": "login",
  "fromSide": "right",
  "toNode": "dashboard",
  "toSide": "left",
  "label": "Sign in",
  "color": "4"
}

id, fromNode, and toNode are required and the node ids must exist. fromSide / toSide are top | right | bottom | left. fromEnd / toEnd are none | arrow; omitted, you get the spec default of plain at the source and an arrow at the target, which is almost always what you want.

Edges render as smooth-step arrows with an optional centered label. Set fromSide / toSide deliberately — an edge left to route itself between two cards on the same row will loop around them.

x-nimbalyst.flow (kind: "flow", with fromElementSelector and trigger) is defined on the edge and preserved through save, but the editor does not yet act on it. Do not describe a board as a click-through prototype on the strength of it.

Worked example

An onboarding flow: three mockup screens wired left to right, a frame around them, and a sticky note holding an open question.

json
{
  "nodes": [
    {
      "id": "frame-flow",
      "type": "group",
      "x": -60,
      "y": -100,
      "width": 2180,
      "height": 640,
      "label": "Onboarding",
      "x-nimbalyst": { "reference": { "kind": "native", "nativeKind": "group" } }
    },
    {
      "id": "signup",
      "type": "file",
      "x": 0,
      "y": 0,
      "width": 640,
      "height": 440,
      "file": "design/mockups/signup.mockup.html",
      "x-nimbalyst": {
        "reference": { "kind": "file", "path": "design/mockups/signup.mockup.html" },
        "label": "Sign up"
      }
    },
    {
      "id": "verify",
      "type": "file",
      "x": 740,
      "y": 0,
      "width": 640,
      "height": 440,
      "file": "design/mockups/verify.mockup.html",
      "x-nimbalyst": {
        "reference": { "kind": "file", "path": "design/mockups/verify.mockup.html" },
        "label": "Verify email"
      }
    },
    {
      "id": "first-doc",
      "type": "file",
      "x": 1480,
      "y": 0,
      "width": 640,
      "height": 440,
      "file": "design/mockups/first-document.mockup.html",
      "x-nimbalyst": {
        "reference": { "kind": "file", "path": "design/mockups/first-document.mockup.html" },
        "label": "First document"
      }
    },
    {
      "id": "note-question",
      "type": "text",
      "x": 740,
      "y": 600,
      "width": 240,
      "height": 180,
      "color": "5",
      "text": "Can verification be deferred until the first share?",
      "x-nimbalyst": { "reference": { "kind": "native", "nativeKind": "sticky" } }
    }
  ],
  "edges": [
    {
      "id": "signup-verify",
      "fromNode": "signup",
      "fromSide": "right",
      "toNode": "verify",
      "toSide": "left",
      "label": "Submit"
    },
    {
      "id": "verify-first-doc",
      "fromNode": "verify",
      "fromSide": "right",
      "toNode": "first-doc",
      "toSide": "left",
      "label": "Confirmed"
    },
    {
      "id": "verify-question",
      "fromNode": "note-question",
      "fromSide": "top",
      "toNode": "verify",
      "toSide": "bottom",
      "toEnd": "none",
      "color": "5"
    }
  ],
  "x-nimbalyst": {
    "version": 1,
    "meta": {
      "name": "Onboarding flow",
      "description": "Sign-up through the user's first document"
    }
  }
}

Checklist before you write the file

  • Every id unique, every fromNode / toNode resolving to a real node
  • Every file card's path exists in the workspace, and matches in both file and reference.path
  • Geometry on the 20px grid, no overlapping cards, gutters left for toolbars
  • Frames earlier in nodes than the cards they contain
  • x-nimbalyst.meta.name set

© nimbalyst, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in packages/extensions/canvas/claude-plugin/skills/canvas of nimbalyst/nimbalyst.

Open the folder on GitHubat commit 6f14adf

Compare with similar skills

Canvas next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Canvas compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Canvas this skillnimbalyst/nimbalyst1.9k—~2.6kAutomated safety check: PassMIT
Obsidian Canvas BoardsAgriciDaniel/claude-obsidian15k—~1.4kAutomated safety check: PassMIT
Canvasopenclaw/openclaw392k—~263Automated safety check: PassMIT
Canva Designerasgeirtj/system_prompts_leaks69k—~2.1kAutomated safety check: PassCC0-1.0
Boardalirezarezvani/claude-skills28k—~610Automated safety check: PassMIT
Hermes Agent Skill AuthoringNousResearch/hermes-agent252k—~3.6kAutomated safety check: PassMIT

Similar skills

  • Obsidian Canvas Boards

    AgriciDaniel/claude-obsidian

    Creates, inspects and updates Obsidian JSON Canvas boards in a vault, with text, file, link, group and edge nodes, using safe recoverable edits.

    15k GitHub stars~1.4k tokensUpdated 27 days ago
    Knowledge ManagementAuto-check passed
  • Canvas

    openclaw/openclaw

    Present hosted widget documents on a connected macOS panel and control panel visibility or navigation.

    392k GitHub stars~263 tokensUpdated today
    Auto-check passed
  • Canva Designer

    asgeirtj/system_prompts_leaks

    Creates and edits Canva designs, generates images and works with brand kits through the canva CLI and Canva's official MCP server, from decks to social posts.

    69k GitHub stars~2.1k tokensUpdated today
    Media & CreativeAuto-check passed
  • Board

    alirezarezvani/claude-skills

    Read, write, and browse the AgentHub message board for agent coordination.

    28k GitHub stars~610 tokensUpdated 1 mo ago
    Agent WorkflowsAuto-check passed
  • Hermes Agent Skill Authoring

    NousResearch/hermes-agent

    Author in-repo SKILL.md files: frontmatter and structure. An agent skill from NousResearch/hermes-agent.

    252k GitHub stars~3.6k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Configuring Oauth2 Authorization Flow

    mukul975/Anthropic-Cybersecurity-Skills

    Configures secure OAuth 2.0 authorization flows, including Authorization Code with PKCE, Client Credentials, and Device Authorization Grant, covering flow selection, PKCE implementation, token…

    34k GitHub stars~1.7k tokensUpdated 1 mo ago
    Backend & APIsAuto-check passed

More from nimbalyst/nimbalyst

All 14 skills in this repo
  • Datamodellm

    nimbalyst/nimbalyst

    Create visual data models for database schemas using Nimbalyst's DataModelLM editor.

    1.9k GitHub stars~713 tokensUpdated today
    Auto-check passed
  • Excalidraw

    nimbalyst/nimbalyst

    Create diagrams and visual drawings using Excalidraw (.excalidraw files).

    1.9k GitHub stars~1.5k tokensUpdated today
    Auto-check passed
  • Extension Development

    nimbalyst/nimbalyst

    Build, install, and hot-reload Nimbalyst extensions using MCP tools.

    1.9k GitHub stars~1.6k tokensUpdated today
    Auto-check passed
  • Git Commit

    nimbalyst/nimbalyst

    Create git commits using Nimbalyst's interactive commit proposal widget.

    1.9k GitHub stars~696 tokensUpdated today
    Auto-check passed
  • Planning

    nimbalyst/nimbalyst

    Create structured plan documents and track work items using YAML frontmatter.

    1.9k GitHub stars~1.1k tokensUpdated today
    Auto-check passed
  • Knowledge Graph

    nimbalyst/nimbalyst

    Write and maintain project wiki pages, decisions, and links.

    1.9k GitHub stars~3.5k tokensUpdated today
    Auto-check passed

Questions about Canvas

What does Canvas do?

Author Nimbalyst Project Canvas boards (.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges. Canvas is an agent skill from nimbalyst/nimbalyst.canvas files) — an infinite canvas whose cards are live editors for real workspace files and shared documents, arranged spatially and wired with edges.

When should I use Canvas?

Canvas fits situations like: the user asks for a canvas; A spatial layout of documents; A flow of screens; A workspace overview.

How do I install Canvas in Claude Code?

Run `npx skills add nimbalyst/nimbalyst --skill canvas -a claude-code`. Or copy the skill folder (packages/extensions/canvas/claude-plugin/skills/canvas in nimbalyst/nimbalyst) into .claude/skills/canvas in your project. Claude Code loads it when a task matches its description.

How do I install Canvas in Codex?

Run `npx skills add nimbalyst/nimbalyst --skill canvas -a codex`. Or copy the skill folder (packages/extensions/canvas/claude-plugin/skills/canvas in nimbalyst/nimbalyst) into .agents/skills/canvas in your project. Codex loads it when a task matches its description.

Can I use Canvas in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add nimbalyst/nimbalyst --skill canvas -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/canvas, .gemini/skills/canvas, .github/skills/canvas and .opencode/skills/canvas in your project.

What does Canvas need to run?

SKILL.md names no scripts, command-line tools or credentials: Canvas is instructions for the agent only.

Does Canvas access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Canvas safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Canvas use?

Canvas is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Canvas use?

About 2.6k tokens (SKILL.md is roughly 10k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Canvas?

Skills that share tags, products or a category with Canvas: Obsidian Canvas Boards (AgriciDaniel/claude-obsidian, 15k stars), Canvas (openclaw/openclaw, 392k stars), Canva Designer (asgeirtj/system_prompts_leaks, 69k stars) and Board (alirezarezvani/claude-skills, 28k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Canvas?

nimbalyst (a GitHub organization) maintains it in nimbalyst/nimbalyst, which has 1,850 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on October 8, 2026.

Source: nimbalyst/nimbalyst on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.