Mokido
Mokido is an MCP server that gives an AI agent read/write access to a live Figma Desktop canvas, with visual verification, component/variable/SVG tools, rollback, and multi-file/job support.
Run arbitrary JavaScript in the Figma sandbox to create, edit, move, style, or delete nodes, with optional PNG capture and helper macros/checkpoints.
Inspect the live document in one call: outlines, props, name/text searches, tables, prototype flows, comparisons, and context.
Take PNG screenshots of nodes or the current selection for visual checks.
Read the live canvas as token-optimized JSX, tree, or JSON with depth and token budgets.
Get selection details and canvas layout/next safe position, including row/grid packing.
Find components/component sets/variants and insert instances with property and text overrides.
Insert raw SVG with scaling, color overrides, and optional component wrapping.
Read Figma Variables collections/modes/tokens and switch variable modes on frames or pages.
Roll back a previous write via checkpoint ID, removing created nodes and restoring modified properties.
Handle long-running code calls with
job_idand blockingfigma_job_status.List connected Figma documents and target a specific open file by fileName.
Provides read and write access to a live Figma canvas, allowing AI agents to create and modify nodes, capture rendered PNGs for visual self-critique, undo changes via checkpoints, and read token-optimized document structure.
⚡ Mokido
MCP server for Figma design. Let an AI agent design in Figma — and actually see what it drew.
Formerly figma-mcp-bridge (@kolganovr/figma-mcp-bridge on npm). Old links redirect here.
A zero-dependency MCP server that gives coding agents read and write access to a live Figma canvas, then closes the loop by handing the rendered PNG back to the model for visual self-critique.

One prompt, two screens: the agent reads your design system, builds them live in the file you have open, checks its own screenshot and fixes the layout. 56 seconds of real time, sped up.
The problem
Coding agents are effectively blind inside Figma. Most Figma MCP servers are read-only —
they flatten a design into text so a model can turn it into code, and the traffic stops there.
The ones that can write are either gated behind a paid Figma seat with a monthly tool-call
quota, or expose a fixed vocabulary of commands (create_rectangle, set_fill) that runs out
the moment the task gets specific.
And none of them answer the question that actually matters after a write: did it look right?
An agent that draws a card, gets back {"ok": true}, and moves on has no way to notice that its
text overflowed, its contrast failed, or its frame landed on top of someone else's work.
This bridge closes that loop, and adds the safety rails an autonomous agent needs to be trusted with a real design file.
Related MCP server: figma-mcp-express
How it's different
This bridge | Figma Dev Mode MCP (official) | |||
Writes to the canvas | ✅ arbitrary JS in the sandbox | ✅ code-to-canvas | ❌ read-only | ✅ fixed command set |
Visual feedback loop | ✅ auto-framed PNG on any write ( | ⚠️ separate screenshot call | ❌ | ❌ |
Undo the agent's work | ✅ | ❌ | n/a | ❌ |
Works on the Figma Free plan | ✅ | ❌ Dev/Full seat, paid plan¹ | ✅ | ✅ |
Tool-call quota | none — it's local | 6 / month on Starter seats¹ | inherits REST rate limits | none |
Token-optimized read of the live doc | ✅ 86–91% smaller² | ⚠️ partial | ❌ REST only | ❌ |
Persistent code modules in-sandbox | ✅ | ❌ | ❌ | ❌ |
Long jobs survive their own timeout | ✅ async | ❌ | n/a | ❌ |
Token budget per call | ✅ capped text, sized images, compact JSON | ❌ | ⚠️ read-side only | ❌ |
Install footprint | 0 npm deps, | Figma desktop + paid seat |
| Bun + a second server process |
The short version: Framelink is the best choice if you only want to turn an existing design into code. The official server is the safest choice if you're already on a paid Figma plan and want Code Connect. This bridge is for the case where the agent is doing the designing — where it needs to write freely, check its own work, and be undoable when it gets it wrong.
How it works
flowchart LR
subgraph AI ["🤖 Coding Agent"]
LLM["Claude Code · Cursor<br/>Antigravity · Windsurf"]
end
subgraph Bridge ["⚡ MCP Server — Node.js, 0 deps"]
Router["Tool Router<br/>+ Job Ledger<br/>+ Target Router"]
Opt["Token Optimizer<br/>REST + LIVE"]
end
subgraph Figma ["🎨 Figma Desktop"]
Plugin["Bridge Plugin<br/>Checkpoint Journal<br/>Component Index"]
Canvas["Live Canvas"]
end
LLM -->|"stdio JSON-RPC"| Router
Router <-->|"WebSocket :8765"| Plugin
Plugin -->|"execute in sandbox"| Canvas
Canvas -->|"export PNG"| Plugin
Plugin -->|"raw node tree"| Opt
Opt -.->|"86% smaller"| LLM
Plugin ==>|"screenshot + warnings"| LLMThe agent speaks plain MCP over stdio. The server owns a hand-rolled RFC 6455 WebSocket on
:8765, which the Figma plugin connects to — so commands reach the canvas with no polling
latency, and results (including multi-megabyte base64 PNGs) stream straight back.
Quickstart
npx mokidoOr, to read the installer before it touches your machine — it's the same script either way, just fetched differently:
git clone https://github.com/kolganovr/mokido.git
cd mokido
node install.mjsBoth copy the server and plugin into place and register the MCP server in every AI client config
they find — Claude Desktop, Claude Code, Cursor, Windsurf, Antigravity. The npm package exists
purely for a shorter first command; it doesn't add a single runtime dependency — dependencies
is empty in its package.json too, and no Python is needed either. node is the only runtime
this project ever needs, for the installer and the server alike.
Then, in Figma Desktop:
Plugins → Development → Import plugin from manifest… → pick
~/.mokido/mcp/figma-plugin/manifest.json(the installer prints the full path; the repo's ownfigma-plugin/has no bridge token)Press
Ctrl + Alt + P(macOS:Cmd + Option + P) to launch MokidoStatus turns green —
CONNECTED
Restart your AI client so it picks up the new tools. Verify anytime with:
npx mokido --doctor # or: node install.mjs --doctorThe live canvas tools need no token. If you also want to read unopened cloud files
(get_file, get_node, get_styles, …), supply a personal access token:
npx mokido --token "your_figma_personal_access_token"
# or: node install.mjs --token "your_figma_personal_access_token"Without a token these 7 tools aren't registered at all — see Tool reference.
Updating
Updating is the same installer run again. It overwrites the copy in ~/.mokido/mcp, keeps your bridge
token and client configs, and removes copies left by older versions.
npx mokido@latest --updateKeep @latest: a bare npx mokido can reuse the version npx cached last time. From a clone:
git pull
node install.mjs --updateThen reopen the plugin in Figma (Ctrl + Alt + P) and restart your AI client. A server that was already
running keeps the old code in memory, and the plugin shows "Update needed" until it restarts.
You don't have to watch for releases: once a day the server checks npm and prints a one-line notice to your client's MCP log when a newer version is out (see the update check). What changed in each version is in CHANGELOG.md.
Agents updating it for you follow the same steps in AGENTS.md.
What it gives the agent
1. Write freely, then look at the result
capture: true returns a PNG of exactly what the call just created or modified — auto-framed
to the changed nodes, never the whole page, and never by hijacking the user's selection.
// agent calls figma_execute_code
{ "code": "const f = figma.createFrame(); /* ... */ return f.id;", "capture": true }
// agent gets back — text + image in one response
{
"ok": true,
"created": ["12:34"],
"warnings": ["Text \"Total\" has ~3.1:1 contrast against its parent fill (WCAG AA wants 4.5:1)."],
"checkpoint_id": "cp_mfk3p2a_7",
"duration_ms": 840
}The warnings array is a cheap auto-lint over the touched subtree — text overflow, low contrast,
zero-size nodes. It catches the obvious mistakes without spending a screenshot round trip.
2. Undo anything the agent did
Every write opens a checkpoint automatically. One call reverses it — no Ctrl+Z rolling back over
the human's own unrelated work.
// agent calls figma_rollback
{ "checkpoint_id": "last" }
// gets back
{
"ok": true,
"checkpoint_id": "cp_mfk3p2a_7",
"label": "Generate checkout flow",
"removed": ["12:34", "12:35"], // nodes the call created — deleted
"restored": ["9:11"], // nodes it modified — properties put back
"missing": [] // ids that no longer exist
}Created nodes are tracked automatically via a Proxy around figma.create*(). Property edits are
tracked when snapshotted. Deletions are honestly reported as unrecoverable rather than silently lost.
3. Read the live canvas for ~700 tokens instead of ~5,000
The same pruning + Pseudo-JSX pipeline that powers the cloud tools, pointed at whatever is open right now. Measured on a realistic 6-card layout: 20 KB of raw API JSON → 2.7 KB (86% smaller).
<Frame id="1:1" name="Landing" w="1440" h="900" row gap="20">
<Frame id="2:0" name="Card" w="300" h="200" col gap="12" pad="20" bg="#FFFFFF" radius="16">
<Icon id="3:0" name="ic_check" size="24" strokeWidth="2" />
<Text id="4:0" color="#1A1A1F" font="Inter 18px">Feature 0</Text>
</Frame>
</Frame>budget_tokens caps the response: if the requested depth overshoots, the server re-serializes the
already-fetched tree shallower — no second round trip — and appends a comment saying what it did
and which id to fetch for more.
4. Teach the sandbox new tricks that survive restarts
Every figma_execute_code call is a fresh function scope, so helpers normally die instantly.
bridge.define compiles and stores a module inside the .fig document:
// once
bridge.define("kit", `
async function label(parent, text) { /* ... */ }
module.exports = { label };
`);
// in any later call — including next week, after a Figma restart
const { label } = bridge.require("kit");5. Long jobs that don't die at the timeout
A generation still running after 45s hands back a job_id instead of failing while the plugin
keeps working. figma_job_status then blocks until the job finishes (up to wait_ms, 45s by
default) — one call, not a polling loop. Progress from progress(step, of, note) is included while
it runs; a job silent for 2 minutes is flagged stalled, and one whose plugin disconnected fails
at once with PLUGIN_DISCONNECTED instead of running out the clock. A call that times out behind a
still-running job says so (PLUGIN_BUSY, naming the job) instead of inviting a blind retry.
6. Built for the token bill
In an agent loop a tool call is paid for twice: once when it returns, and again on every later turn, because its output stays in the context that each turn re-reads. So the bridge optimizes for fewer turns and smaller residue:
Default | Why | |
Screenshots |
| image tokens scale with pixel area (~w·h/750); each image is described as |
| off | the auto-lint |
Text results | compact JSON; every response is kept under | some clients (Antigravity) spill output over ~4 KB into a file the model must open with an extra call; the full value stays in |
Cheap reads |
| one-line-per-node outlines and pass/fail lists instead of one hand-written dump per call |
Plugin reloads | calls wait up to 8s for the plugin to reconnect | a failure costs a turn to read and another to retry |
Server instructions | ≈1.9k chars, economy rules first | some clients truncate instructions at ~2000 chars |
Every default is overridable per call, and per install through env: FIGMA_MCP_SCALE,
FIGMA_MCP_MAX_PX, FIGMA_MCP_MAX_IMAGES, FIGMA_MCP_MAX_OUTPUT_BYTES, FIGMA_MCP_MAX_OUTPUT_CEILING (0 = no ceiling), FIGMA_MCP_JOB_WAIT_MS,
FIGMA_MCP_ESCALATE_MS, FIGMA_MCP_RECONNECT_GRACE_MS.
What's new
4.4.0: a new plugin window with the Moki companion (12 states, compact square/bar view, settings with versions,
dark theme, English/Russian); every client now shares one install folder, ~/.mokido/mcp, and a server exits with
its session, so a stale one no longer holds port 8765; figma_inspect reads property definitions of component
sets, lists all texts of a frame (props: ["texts"]) and names the page when find has no match.
Re-import the plugin once from ~/.mokido/mcp/figma-plugin/manifest.json. History: CHANGELOG.md.
4.2.7: compare / like diffs end with a REF line of exact reference values, so the agent fixes copies
without an extra props read; find returns the subtree of what it found; view: "flow" works without
node_ids; bridge.setText keeps mixed fonts. On the internal bench (Gemini Flash, seeded tasks) model turns
per series went 112 -> 83 and verify turns after a write 11 -> 1. Reload the Figma plugin after updating.
History: CHANGELOG.md.
Tool reference
Tools are served in tiers, so the schema list sent to the model on every turn stays proportional to what's actually usable:
Tier | Count | Registered when |
Core | 8 | always |
Extended | 5 | always |
REST | 7 | only with |
Legacy | 3 | only with |
Without a REST token an agent sees 13 tools instead of 23 — and never wastes a call on
something that would only return REST_TOKEN_MISSING.
Core — live canvas
Tool | Description |
| Run JS in the Figma sandbox. Injects |
| Read many live nodes in one call: compact outline (or |
| Token-optimized read of the live document ( |
| PNG of specific |
| Cached, tokenized, fuzzy component search — variants, properties, keys. |
| Instantiate a component/variant, apply text overrides, place into AutoLayout. |
| Insert raw SVG with proportional scaling, recoloring, optional component wrapping. |
| Variable collections, modes, and token values. |
| Undo a previous write call's checkpoint. |
Extended — live canvas
Tool | Description |
| Geometry, compact hex fills, parent/page, AutoLayout context of the selection. |
| Artboard bounds + a safe |
| Switch theme mode (Dark/Light/Brand) on a frame or page. |
| Wait (blocking, |
| List connected Figma documents for multi-file targeting. |
REST — Figma Cloud (needs a token)
Tool | Description |
| Token-optimized cloud file/subtree. Supports |
| Render nodes to PNG/SVG/PDF via Figma's renderer. |
| Published styles and design-system components. |
| Read and post file comments. |
Legacy — opt-in via FIGMA_MCP_LEGACY_TOOLS=1
figma_create_ui_card · get_me · get_image_fills
Engineering notes
The parts that were harder than they look — and why the code is shaped the way it is.
The obvious way to persist helpers between calls is "stash the source, eval it next time." It
fails silently: eval is a bound function in Figma's sandbox, which by spec makes every call
an indirect eval — declarations inside it reach neither the caller's scope nor globalThis.
Nothing is defined, nothing throws.
bridge.define is built on new Function instead, whose bodies are ordinary function scopes.
The runtime's bridge.info() reports this contract to the agent on request, so it can ask instead
of guessing.
Zero npm dependencies isn't a vanity metric here — it means git clone && node install.mjs works
on a locked-down machine with no registry access, and there's no supply chain to audit for something
that executes arbitrary JS inside your design files.
The cost is owning the framing: masking, fragmented continuation frames (a 4 MB screenshot arrives split, and treating each fragment as a whole message silently dropped it until the tool call timed out 40s later), ping/pong liveness, and a 64 MB ceiling so one frame can't exhaust memory.
Figma caps pluginData entries at roughly 100 KB — measured in bytes. The original chunker
sliced by JS string length, so a module written in Cyrillic (~2 bytes/char) produced "60,000-char"
chunks that were really 120 KB, and setPluginData threw. The splitter now walks the real UTF-8
budget and never tears a surrogate pair.
Every agent spawns its own copy of the server; the first to bind :8765 owns the plugin socket and
the rest proxy to it. When the owner exits, a proxy has to be able to take over — otherwise every
surviving agent stays permanently broken until restarted. A 5-second watchdog retries the bind, and
a failed proxy call triggers an immediate takeover attempt.
The original collision engine rescanned every top-level node for each of up to 200 candidate positions, then did it again in a second function in the same tick — and only ever stepped along one axis, so 20 generated screens became a mile-long ribbon nobody could zoom out to see.
Bounds are now computed once and shared; collisions go through a 500px grid hash; and layout:"grid"
shelf-packs into a compact rectangle.
Raw Plugin API errors are famously unhelpful. Failures are matched against known modes and rewritten
with a HINT: line naming the API that actually works, plus a stable machine-readable code
(FONT_NOT_LOADED, INSTANCE_TRANSFORM_LOCKED, STALE_NODE_ID, AUTOLAYOUT_HUG_RESIZE,
AMBIGUOUS_TARGET, …) so agents and tooling can branch on the failure type without parsing prose.
:8765 is loopback, but loopback is reachable by any web page the user happens to have open.
Two independent gates: an Origin allowlist (a browser cannot forge Origin, so a page on
evil.com is rejected at the handshake) and a shared token that install.mjs generates and bakes
into both the MCP config and the installed plugin — which also covers the sandboxed-iframe case,
where a hostile page can present Origin: "null" too.
Running straight from a clone with no token, the Origin gate still applies and the server prints a warning, so it degrades rather than silently opening up.
On startup an installed server (the copy install.mjs wrote, marked by a version.json beside the
code) compares its own version with the latest mokido release on npm via one registry request,
and prints a one-line notice to stderr if a newer release is out. It compares releases, not commits,
so it works the same for npx and clone installs and stays quiet about unreleased work on main.
It never applies anything itself — updating is still a manual, deliberate step.
This is the one thing in this project that assumes network access, so it's built to disappear
cleanly when there isn't any: the check is fired without being awaited (never delays initialize
or the first tool call), a failed/slow/offline request is caught and silently skipped, and results
are cached for 24h so it doesn't hit the registry on every launch or run once per spawned
copy of the server. Set FIGMA_MCP_NO_UPDATE_CHECK=1 to turn it off entirely — worth doing on the
locked-down machines the previous note is about.
Testing
Seven dependency-free suites, all runnable with bare node:
node tests/bridge-runtime.test.js # sandbox runtime, module persistence, checkpoint/rollback
node tests/layout-packer.test.js # row/grid packing, collision grid
node tests/optimizer.test.js # jsx/tree/json serialization, budget truncation
node tests/mcp-protocol.test.js # real server over stdio: initialize, tools/list, tiering
node tests/install.test.js # config merge/reuse, token persistence, stale-file cleanup
node tests/token-economy.test.js # fake plugin over a real WebSocket: budgets, long-poll, proxy
node tests/output-shrink.test.js # line-wise shrinking, offset paging, output ceiling, syntax pre-checkmcp-protocol.test.js spawns the actual server as a child process and speaks NDJSON to it — the
same transport a real MCP client uses — rather than importing internals.
Repository layout
mokido/
├── figma/ # MCP server (Node.js, stdio + WebSocket)
│ ├── index.js # protocol, tool router, job ledger, target router
│ ├── optimizer/ # AST pruner, style collapser, JSX/tree serializers
├── figma-plugin/ # Figma Desktop plugin
│ ├── code.js # sandbox executor, bridge runtime, checkpoints, capture
│ └── ui.html # HUD — stream, settings, control (pause / undo)
├── docs/GUIDE.md # long-form agent guide (server instructions are inline in index.js)
├── tests/ # 7 suites, 0 dependencies
├── install.mjs # cross-platform installer, updater, doctor (Node only)
└── AGENTS.md # onboarding protocol for AI agentsSources
Seat and quota requirements for the official server — Figma: Guide to the Figma MCP server, Figma Developer Docs: Rate limits & access
Token reduction measured on a 6-card layout fixture that mimics real REST output: 20,587 B raw → 2,862 B Pseudo-JSX (86.1%) / 1,912 B tree (90.7%). The fixture is checked in and asserted — run
node tests/optimizer.test.jsto reproduce the exact figures.
Comparison reflects publicly documented behaviour as of August 2026. Alternatives are actively developed — verify current capabilities before making a decision on this table alone.
License
MIT · Built by Roman Kolganov
Available Tools
14 toolsfigma_execute_codeADestructive
WRITE tool: run JavaScript in the open Figma document (Figma Desktop + 'Mokido' plugin) to create, edit, move, style or delete nodes. To READ (explore a section, list children, find layers, check texts / fonts / variants / table columns) call figma_inspect instead: every id in one call, one line per node, view:"table" for tables; a read script here costs a model turn per question and its reply names the figma_inspect call to use (use_instead). Every call is a full model turn, so make the whole change in ONE call — all screens / breakpoints, verified in the same code with bridge.check — and put capture_node_ids on that same call instead of a separate figma_screenshot. Globals: figma, await ensureFont(family, style), getFreePosition(w, h), bridge: bridge.check({ '1:2': { width: 320, fill: '#FFFFFF' } }) -> pass/fail list; bridge.define(name, src) + bridge.require(name) keep helpers between calls (each call is a fresh function body — never eval); return bridge.info() lists the rest. Macros: bridge.replaceWithInstance(target, comp, {props, text}), bridge.setProps(inst, {Status: 'Dropdown'}), bridge.setText(root, {layer: '…'}) or bridge.setText(textNode, 'new text') (keeps mixed font styles, sizes, colours by words: THE way to replace the text of a label with several font styles — never characters =, which flattens them, and never a per-character getRangeFontName/setRangeFontName loop), bridge.shift(ids, {dx, dy}), bridge.moveInto(ids, section, {layout, gap}), bridge.fitSection(s), bridge.context(id), bridge.componentize(nodeObject) -> new COMPONENT, same name/size/look, NEW id (never raw createComponentFromNode), prototype links: await bridge.openOverlay(triggerId, overlayFrameId, { position: 'BOTTOM_CENTER', dim: 0.4, closeOnOutside: true }) (the overlay look is read-only in the Plugin API: the reply names the MANUAL STEP for the user), bridge.closeOverlay(id), bridge.setReaction(id, { to: frameId, nav: 'NAVIGATE' }); read links with figma_inspect view:"flow". Call macros directly, don't read their source. Never findAll the whole page — scope to a section. Return only the ids/flags you need: responses over max_output_bytes (3500) are shrunk, the full value stays in bridge.state.lastResult.
Inputs: code (required), diff, async, scale, +6 more.
figma_find_componentsARead-onlyIdempotent
Find and inspect components, component sets, variants, and component property definitions in the active Figma file without flooding LLM context. Returns names, variant values, keys, and IDs.
Inputs: limit, query, target, page_name, +2 more.
figma_get_canvas_layoutARead-onlyIdempotent
Inspect top-level frames and artboards on the active Figma page to prevent overlap. Returns all screen coordinates, bounding box, and a calculated 'suggestedNextPosition' for placing new artboards safely.
Inputs: gap, limit, layout, target, +2 more.
figma_get_selectionARead-onlyIdempotent
Get information and properties (dimensions, coordinates, text, fills, parent, page, AutoLayout context) of the currently selected nodes on the Figma canvas.
Inputs: target.
figma_get_variablesARead-onlyIdempotent
Retrieve all Figma Variable collections, modes (e.g. Light/Dark), and design tokens (colors, numbers, strings, booleans) from the active document.
Inputs: limit, target, collection_name.
figma_insert_component_instanceA
Create and insert an instance of a master component or component set into the canvas or target AutoLayout container. Supports selecting variants, applying text overrides with auto font loading, and can return a PNG screenshot (capture: true) for visual verification.
Inputs: scale, target, capture, position, +5 more.
figma_insert_svgA
Insert raw SVG/vector code directly into Figma canvas or target AutoLayout container with automatic scale-proportional resizing, fill/stroke color overrides, optional component creation, and an optional PNG screenshot (capture: true).
Inputs: svg_code (required), name, scale, width, +9 more.
figma_inspectARead-onlyIdempotent
READ the live Figma document in ONE call — the tool for any read (not figma_execute_code); pass every node id you need at once instead of one call per node. Default: a compact outline, one line per node (TYPE "name" #id WxH @x,y [V gap8 pad16 fill/hug a:start/center] size:fill/hug fill:#FFF stroke:#E8EAF0/0,1,1,1 r0,0,3,3 shadow:… font "text…"; non-defaults only; a TEXT named after its text drops the name; a:P/C = primaryAxisAlignItems/counterAxisAlignItems, start = MIN, center = CENTER, end = MAX, between = SPACE_BETWEEN, absent = MIN/MIN), as deep as fits the reply unless depth is given; children:N marks a node with more below. like: "<ref id>" (or several) FINDS every copy of the reference under node_ids (screens/sections; default: its section) — instances of the same component, or same name — and diffs each against it: the call for "redo these like the reference" when you don't know where the copies are. compare: "<ref id>" lists, per node_id, every difference from that reference (layout, sizing, stroke, radius, effects, clip, font, variant; children matched by name) and ends with REF lines, the reference's exact values to copy into the write (no props read needed) — one call for "make these like that one", and the check after the write. props returns exact values instead: { id: { prop: value } } (also fill, stroke, text, font, layout, padding, parent, children, main, variant, props, absolute, reactions, connector). find searches node names under node_ids (or the current page); find_text searches text content. view: "map" = canvas map (sections, breakpoints). view: "flow" = the prototype links under node_ids (a section or screens): per screen, source, trigger, action, target, transition, plus the overlay settings of overlay targets — never scan the page for reactions. context: true adds ancestor sections, breakpoints and component variants. offset skips the first N outline lines when a reply ends with pass offset=K. check returns only mismatches. Combine them freely. Changes nothing.
Inputs: ids, find, like, view, +13 more.
figma_job_statusARead-only
Wait for a figma_execute_code call that came back as { status: 'running', job_id } (it ran past 45s, or was called with async: true). BLOCKS until the job finishes or wait_ms passes, then returns the same result/screenshot the synchronous call would have — so call it once, not in a polling loop. stalled: true means the plugin looks frozen: STOP, ask the user to check Figma; a second figma_job_status for that job returns at once with the same advice. A finished job is forgotten after being read once.
Inputs: job_id (required), wait_ms.
figma_list_targetsARead-onlyIdempotent
List every Figma document currently connected to this bridge (fileName, current page, which one is focused). Use this when a LIVE tool call fails with AMBIGUOUS_TARGET, or before working with a specific file among several open at once — pass the fileName as target on any LIVE tool.
Inputs: no named parameters.
figma_read_canvasARead-onlyIdempotent
Read the LIVE, currently-open Figma document as token-optimized Pseudo-JSX/Tree/JSON — the same pipeline get_file/get_node use for the cloud API, applied to whatever is open in Figma Desktop right now. Prefer this over hand-writing a tree walk in figma_execute_code; it is far cheaper in tokens and its output format matches get_file/get_node exactly.
Inputs: depth, format, target, node_ids, +2 more.
figma_rollbackADestructive
Undo what a previous write call (figma_execute_code, figma_insert_component_instance, figma_insert_svg) did: removes nodes it created and restores properties on nodes it modified. Deletions the code performed on pre-existing nodes are never recoverable. Every write call's response includes the checkpoint_id to pass here.
Inputs: target, checkpoint_id.
figma_screenshotARead-onlyIdempotent
Capture PNG screenshots of specific nodes or the current selection in Figma, to check layout, contrast, typography and spacing. Each image costs ~1k tokens at the default max_px, and stays in context: check mechanical things through write-call warnings or bridge.check first, and take one screenshot per finished stage rather than after every edit. Right after a figma_execute_code edit, put capture_node_ids on that call instead — this tool is then an extra turn.
Inputs: scale, max_px, target, node_ids, +1 more.
figma_set_variables_modeAIdempotent
Switch the active Figma Variables mode (e.g. Dark Mode, Light Mode, Brand theme) for a specific artboard/frame or the entire page.
Inputs: mode_name (required), collection_name (required), scale, target, +2 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 14 tools
Write-side tools are fairly distinct, but the read surface has several overlapping options: figma_inspect, figma_read_canvas, figma_get_selection, figma_get_canvas_layout, and figma_find_components all retrieve live document information. Descriptions give guidance, but an agent may still struggle to choose between figma_inspect and figma_read_canvas for general document reading.
All tools use a consistent figma_ snake_case prefix, and most follow a verb_noun pattern (get_selection, find_components, insert_svg, set_variables_mode). Minor deviations exist in noun-style names like figma_screenshot and figma_job_status, and verb-only names like figma_inspect and figma_rollback.
At 14 tools, the set is within the ideal range and each tool has a recognizable role in the Figma bridge workflow. The count is well-scoped despite some tools being very broad.
The surface covers reading, writing, undo, async job handling, screenshots, component insertion, SVG insertion, selection inspection, and variables mode switching. Minor gaps remain around creating/editing Figma Variables and cloud/API document tools, but figma_execute_code can cover many custom write operations.
Maintenance
Related MCP Connectors
Live React design-system APIs, patterns, and code validation so AI agents build real UI, not slop.
Create and edit architecture diagrams from your AI agent; get an SVG and a live editable canvas.
Design canvas for agents: create, edit, render and export Figma-shaped designs over MCP.
Lets your AI build and edit websites in Xodia, a visual website builder for real code (Vite, React, Tailwind CSS or plain HTML in your own git repo): read and write site files, create pages and CMS collections, run typecheck, design and link checks, take screenshots, manage forms and translations, and pick up change requests the user sends from the visual editor. Sign in once with your Xodia account (OAuth). Free plan, no card needed. Setup: https://xodia.ai/ai/mcp
Related MCP Servers
- AlicenseAqualityCmaintenanceEnables AI agents to read and draw on Figma canvas safely, with built-in safeguards for layout, overlays, and verification.5107MIT
- AlicenseNot gradedqualityBmaintenanceFast, quota-free Figma automation for coding agents, enabling AI agents to read, edit, migrate, audit, screenshot, and prototype inside Figma Desktop files.9MIT
- AlicenseNot gradedqualityDmaintenanceEnables MCP-compatible AI clients to read and modify the user's currently open Figma file by executing JavaScript in Figma's sandbox, all through a local bridge with status monitoring, node jumping, and automatic rollback on errors.3MIT
- AlicenseNot gradedqualityCmaintenanceEnables a locally run agent or script to read and write a real Figma layer tree through a local bridge — inspecting node structure and variables, exporting screenshots into the model's context, and performing batched create/update/move/rename/delete operations that collapse into a single undo step. All traffic stays on localhost between the bridge and the Figma plugin, with design-token auditing, binding, and creation included.MIT