Available Tools - Detailed Documentation
This guide provides detailed documentation for each tool, including when to use them and best practices.Note: Local Mode (NPX/Git) provides 114 tools with full read/write capabilities and real-time monitoring. Remote Mode provides 9 read-only tools by default, or 101 tools (including full write access) when paired with the Desktop Bridge plugin via Cloud Relay. Tools marked “Local” in the table below require Local Mode. Tools marked “Local / Cloud” work in both Local Mode and Cloud Mode (after pairing).
Quick Reference
🧭 Navigation & Status Tools
figma_navigate
Switch the active Figma file target (Local Mode) or navigate the cloud headless browser to a file (Remote/Cloud Mode).
Usage:
lock (optional, Local Mode): Pass lock: true to pin the target file. Once pinned, new plugin connections, reconnects, and your own selection/page changes in other files will not move the command target — so an AI agent can work in one file while you work in another without writes routing to the wrong file. The pin auto-releases when the pinned file’s plugin disconnects or navigates to a different file; switching to another file (or passing lock: false) also releases it. Use figma_list_open_files to check the current targetLocked state before a write batch.
Remote/Cloud Mode: Navigates the Cloudflare-hosted headless browser to the URL and starts monitoring.
Returns:
- Navigation status
- Current URL
- Connection or monitoring status
locked— whether the target is now pinned (Local Mode)
figma_get_status
Check connection and monitoring status. In local mode, validates WebSocket transport connectivity and shows connection state.
Usage:
- Setup validation (local mode only):
setup.valid- Whether the WebSocket transport is availablesetup.message- Human-readable statussetup.transport- Transport status (websocketornone)setup.setupInstructions- Step-by-step setup guide (if no transport available)setup.ai_instruction- Guidance for AI assistants
- Browser connection status
- Console monitoring active/inactive
- Current URL (if navigated)
- Number of captured console logs
- Call this tool first when starting a session in local mode
- If
setup.validis false, guide user to install and run the Desktop Bridge Plugin
📋 Console Tools (Plugin Debugging)
figma_get_console_logs
💡 Plugin Developers in Local Mode: This tool works immediately - no navigation required! Just check logs, run your plugin in Figma Desktop, check logs again. AllRetrieve console logs with filters. Usage:[Main],[Swapper], etc. plugin logs appear instantly.
count(optional): Number of recent logs to retrieve (default: 100)level(optional): Filter by log level (default: ‘all’)since(optional): Unix timestamp in milliseconds - only logs after this time
- Array of console log entries with:
timestamp: Unix timestamp (ms)level: ‘log’, ‘info’, ‘warn’, ‘error’, ‘debug’message: The log messageargs: Additional arguments passed to console methodstackTrace: Stack trace (for errors)
figma_watch_console
Stream console logs in real-time for a specified duration.
Usage:
duration(optional): How long to watch in seconds (default: 30, max: 300)level(optional): Filter by log level (default: ‘all’)
- Real-time stream of console logs captured during the watch period
- Summary of total logs captured by level
figma_clear_console
Clear the console log buffer.
Usage:
- Confirmation of buffer cleared
- Number of logs that were cleared
🔍 Debugging Tools
figma_take_screenshot
Capture screenshots of Figma UI.
Usage:
target(optional): What to screenshot'plugin': Just the plugin UI (default)'full-page': Entire scrollable page'viewport': Current visible viewport
format(optional): Image format (default: ‘png’)quality(optional): JPEG quality 0-100 (default: 90)filename(optional): Custom filename
- Screenshot image
- Metadata (dimensions, format, size)
figma_reload_plugin
Reload the current Figma page.
Usage:
- Reload status
- New page URL (if changed)
🔁 Token Sync Tools
Bidirectional design token synchronization between Figma variables and your codebase — replaces Style Dictionary and Tokens Studio’s export pipeline for popular styling methods. The canonical pivot format is DTCG JSON (W3C Design Tokens Community Group spec), in your choice of dialect: the legacy hex-string form (default) or the DTCG 2025.10 object form. All ten output formats listed below are fully implemented. Import applies the complete diff plan back to Figma — value updates, creates, renames, alias writes, and (underreplace) deletes.
figma_export_tokens
Export Figma variables to design token files in your codebase. Pulls every variable across every collection and mode, normalizes to the internal token model, and fans out to one or more output formats.
Usage (zero-arg with tokens.config.json):
DTCG dialect (
dtcgDialect):
Import accepts both dialects unconditionally — no flag needed on
figma_import_tokens, and mixed-dialect token files diff correctly (object colors compare equal to their hex equivalents, { value: 16, unit: "px" } equals 16).
Diff-aware merge: Default strategy: "merge" only writes files whose content actually changed. Use strategy: "dry-run" to preview without writing. Use strategy: "replace" to wipe and rewrite.
Round-trip safety: Every exported token carries its Figma variableId and collectionId in DTCG $extensions["figma-console-mcp"]. Renames on either side don’t create duplicates — the ID is the primary match key. Also stamps lastSyncedValue (per-mode snapshot) and lastSyncedAt so two-sided conflicts can be detected on import, plus variable scopes (omitted when default) and per-platform codeSyntax so metadata survives the round-trip.
Cloud Mode: Omit configPath and outputPath. The tool returns token content inline in the response; have your AI client write the files via its own Edit/Write tools. File I/O (autodiscovery, automatic writes) is Local Mode only.
Cross-library aliases: Variables that reference targets in a published library (not in this file’s local variable set) get stamped with {__library:VariableID:...} references — the original Figma ID is preserved for round-trip. CSS formatters emit a comment for skipped tokens with the original library variable ID, so you can see exactly what’s missing and decide how to handle it.
figma_import_tokens
Push code-side token edits back to Figma. Parses any supported source format, diffs against current Figma state, and applies only the deltas via the Plugin API.
Usage (zero-arg with tokens.config.json):
Match priority (how a code-side token gets paired to a Figma variable):
- Figma variable ID stored in
$extensions["figma-console-mcp"].variableId. Survives renames. - Token path (e.g.
color.primary). Used when metadata is absent (first sync, hand-authored DTCG). - Value fingerprint. Used to detect no-op writes — same hash means no API call.
onConflict: "ask" (default) surfaces the conflict and writes nothing. Use "figma-wins" / "code-wins" to auto-resolve, or "skip" to leave conflicts alone and proceed with the rest.
What the apply phase does:
- ✅
toUpdate(value changes on existing variables): applied via the plugin bridge (figma.variables.setValueForMode). Multi-mode supported. Renames matched by round-trip variable ID apply as name changes on the existing variable — a token-path rename never becomes a create+delete pair.scopesandcodeSyntaxmetadata changes apply too (a code-side absent field means “no opinion” and never resets Figma metadata). - ✅
toCreate(new variables and collections): missing collections are created with their full mode lists; missing variables are created with inferred or round-trip-recorded types, values set across all modes in dependency order — literal values first, alias values in a second pass so their targets exist. TIMING/EASING tokens are skipped with a clear warning (the Plugin API cannot create those types). - ✅ Alias updates (code-side
{color.primary}references): written as real{ type: "VARIABLE_ALIAS", id }values. Resolver priority: just-created variable → live Figma snapshot → pending in this batch → recorded$extensionsvariable ID. Unresolvable references are skipped with a warning. - ✅
toDelete(Figma-only variables): applied only understrategy: "replace", and announced loudly in the response. The defaultmergestrategy reports them without deleting.
applyResult.errors[].
Cloud Mode: Pass tokens inline via payload (single file) or files (multi-file). Omit configPath. The apply phase works in Cloud Mode because it routes through the paired Desktop Bridge plugin via the Cloud Plugin Relay — transport-agnostic.
🎨 Design System Tools
⚠️ All Design System tools require FIGMA_ACCESS_TOKEN configured in your MCP client.
See Installation Guide for setup instructions.
figma_get_variables
Extract design tokens/variables from a Figma file. Supports both main files and branches.
Usage:
figma_navigate, you can omit fileUrl entirely:
fileUrl(optional): Figma file URL - supports main files and branches (uses current if navigated)includePublished(optional): Include published variables (default: true)enrich(optional): Add exports and usage analysis (default: false)export_formats(optional): Code formats to generateinclude_usage(optional): Include usage in styles/componentsinclude_dependencies(optional): Include dependency graphrefreshCache(optional): Force fresh data fetch, bypassing cache
- Variable collections
- Variables with modes and values
- Summary statistics
- Export code (if
enrich: true) - Usage information (if
include_usage: true) - Branch info (when using branch URL):
fileKey,branchId,isBranch
figma_get_styles
Get all styles (color, text, effects) from a Figma file.
Usage:
fileUrl(optional): Figma file URLenrich(optional): Add exports and usage (default: false)export_formats(optional): Code formats to generateinclude_usage(optional): Show where styles are usedinclude_exports(optional): Include code examples
- All styles (color, text, effect, grid)
- Style metadata and properties
- Export code (if
enrich: true) - Usage information (if requested)
figma_get_component
Get component data in two export formats: metadata (default) or reconstruction specification.
Usage:
fileUrl(optional): Figma file URLnodeId(required): Component node ID (e.g., ‘123:456’)format(optional): Export format -'metadata'(default) or'reconstruction'enrich(optional): Add quality metrics (default: false, only for metadata format)
- Component metadata and documentation
- Properties and variants
- Bounds and layout info
- Token coverage (if
enrich: true) - Use for: Documentation, style guides, design system references
- Complete node tree specification
- All visual properties (fills, strokes, effects)
- Layout properties (auto-layout, padding, spacing)
- Text properties with font information
- Color values in 0-1 normalized RGB format
- Validation of spec against plugin requirements
- Use for: Programmatic component creation, version control, component migration
- Compatible with: Figma Component Reconstructor plugin
figma_get_component_for_development
Get component data optimized for UI implementation, with visual reference.
Usage:
fileUrl(optional): Figma file URLnodeId(required): Component node IDincludeImage(optional): Include rendered image (default: true)
- Component image (rendered at 2x scale)
- Filtered component data with:
- Layout properties (auto-layout, padding, spacing)
- Visual properties (fills, strokes, effects)
- Typography
- Component properties and variants
- Bounds and positioning
figma_get_component_image
Render a component as an image only.
Usage:
fileUrl(optional): Figma file URLnodeId(required): Node ID to renderscale(optional): Scale factor (default: 2)format(optional): Image format (default: ‘png’)
- Image URL (expires after 30 days)
- Image metadata
figma_get_file_data
Get file structure with verbosity control.
Usage:
fileUrl(optional): Figma file URLdepth(optional): Depth of children tree (max: 3)verbosity(optional): Data detail level'summary': IDs, names, types only (~90% smaller)'standard': Essential properties (~50% smaller)'full': Everything
nodeIds(optional): Retrieve specific nodes onlyenrich(optional): Add statistics and metrics
- File metadata
- Document tree (filtered by verbosity)
- Component/style counts
- Statistics (if
enrich: true)
figma_get_file_for_plugin
Get file data optimized for plugin development.
Usage:
fileUrl(optional): Figma file URLdepth(optional): Depth of children (max: 5, default: 2)nodeIds(optional): Specific nodes only
- Filtered file data with:
- IDs, names, types
- Plugin data (pluginData, sharedPluginData)
- Component relationships
- Lightweight bounds
- Structure for navigation
Tool Comparison
When to Use Each Tool
For Component Development:figma_get_component_for_development- Best for implementing UI components (includes image + layout data)figma_get_component_image- Just need a visual referencefigma_get_component- Need full component metadata
figma_get_file_for_plugin- Optimized file structure for pluginsfigma_get_console_logs- Debug plugin codefigma_watch_console- Monitor plugin execution
figma_get_variables- Design tokens with code exportsfigma_get_styles- Traditional styles with code exportsfigma_get_file_data- Full file structure with verbosity control
figma_get_console_logs- Retrieve specific logsfigma_watch_console- Live monitoringfigma_take_screenshot- Visual debuggingfigma_get_status- Check connection health
✏️ Design Creation Tools
⚠️ Requires Desktop Bridge Plugin: These tools require the Desktop Bridge plugin running in Figma. In Local Mode, the plugin connects via WebSocket. In Cloud Mode, pair first using figma_pair_plugin to connect through the cloud relay.
figma_execute
The Power Tool - Execute any Figma Plugin API code to create designs, modify elements, or perform complex operations.
When to Use:
- Creating UI components (buttons, cards, modals, notifications)
- Building frames with auto-layout
- Adding text with specific fonts and styles
- Creating shapes (rectangles, ellipses, vectors)
- Applying effects, fills, and strokes
- Creating pages or organizing layers
- Any operation that requires the full Figma Plugin API
code(required): JavaScript code to execute. Has access tofigmaglobal object.timeout(optional): Execution timeout in ms (default: 5000, max: 30000)fileKey(optional, Local Mode only): Run against a specific connected file instead of the active one, without changing the active file or releasing target lock. Get connected fileKeys fromfigma_list_open_files. Cloud Mode pairs with a single plugin instance and rejects this parameter rather than silently running against the paired file.
- Whatever the code returns (use
returnstatement) - Execution success/failure status
fileContext— the file name and key as reported by the plugin that ran the code, so you can confirm it executed where you intended
- Always use
awaitfor async operations (loadFontAsync, getNodeByIdAsync) - Return useful data (node IDs, names) for follow-up operations
- Position elements relative to viewport center for visibility
- Select created elements so users can see them immediately
- Use try/catch for error handling in complex operations
figma_execute_across_files
Local Mode only. Run the same code in several connected files at once, concurrently. Built for cross-file work on a multi-file design system — auditing every file for the same problem, or applying the same fix to a set of them — instead of switching the active file and running figma_execute once per file.
Each file’s code runs in that file’s own plugin context, so one file’s failure or timeout doesn’t affect the others. Results come back as a per-file map.
When to Use:
- Checking the same thing across a library split over multiple files (“which files still use the old text styles?”)
- Applying one mechanical fix across a known set of files
- Any read that you’d otherwise repeat file by file
code(required): JavaScript to run in each targeted file. Samefigmaglobal asfigma_execute.fileKeys(optional): Which connected files to target. Get them fromfigma_list_open_files.allFiles(optional, defaultfalse): Target every connected file.timeout(optional): Per-file timeout in ms (default: 10000, max: 30000). Applied independently per file — one unresponsive file doesn’t delay the rest.
fileKeys or allFiles: true. The tool refuses to run otherwise. This is deliberate: allFiles executes your code in files you may be actively editing, including one pinned by target lock, so hitting everything is a decision rather than what happens when you leave a parameter out. Name the files explicitly for anything that writes.
Returns:
fileContextis reported by the plugin that actually ran the code — use it to confirm each result came from the file you addressed.missingFileKeyslists requested files that aren’t currently connected; the rest still run.- The response is only marked as an error if every targeted file failed.
🔧 Variable Management Tools
⚠️ Requires Desktop Bridge Plugin: These tools require the Desktop Bridge plugin running in Figma. In Local Mode, the plugin connects via WebSocket. In Cloud Mode, pair first using figma_pair_plugin to connect through the cloud relay.
figma_create_variable_collection
Create a new variable collection with optional modes.
When to Use:
- Setting up a new design system
- Creating themed variable sets (colors, spacing, typography)
- Organizing variables into logical groups
name(required): Collection nameinitialModeName(optional): Name for the default mode (otherwise “Mode 1”)additionalModes(optional): Array of additional mode names to create
- Created collection with ID, name, modes, and mode IDs
figma_create_variable
Create a new variable in a collection.
When to Use:
- Adding design tokens to your system
- Creating colors, spacing values, text strings, or boolean flags
- Setting up multi-mode variable values
name(required): Variable name (use/for grouping)collectionId(required): Target collection IDresolvedType(required):"COLOR","FLOAT","STRING", or"BOOLEAN"valuesByMode(optional): Object mapping mode IDs to valuesdescription(optional): Variable descriptionscopes(optional): Where variable can be applied
- COLOR: Hex string
"#FF0000"or"#FF0000FF"(with alpha) - FLOAT: Number
16or1.5 - STRING: Text
"Hello World" - BOOLEAN:
trueorfalse
figma_update_variable
Update a variable’s value in a specific mode.
When to Use:
- Changing existing token values
- Updating theme-specific values
- Modifying design system tokens
variableId(required): Variable ID to updatemodeId(required): Mode ID to update value invalue(required): New value (format depends on variable type)
figma_rename_variable
Rename a variable while preserving all its values.
When to Use:
- Reorganizing variable naming conventions
- Fixing typos in variable names
- Moving variables to different groups
variableId(required): Variable ID to renamenewName(required): New name (can include/for grouping)
figma_delete_variable
Delete a variable.
When to Use:
- Removing unused tokens
- Cleaning up design system
- Removing deprecated variables
figma_delete_variable_collection
Delete a collection and ALL its variables.
When to Use:
- Removing entire token sets
- Cleaning up unused collections
- Resetting design system sections
figma_add_mode
Add a new mode to an existing collection.
When to Use:
- Adding theme variants (Dark mode, High Contrast)
- Adding responsive breakpoints (Mobile, Tablet, Desktop)
- Adding brand variants
collectionId(required): Collection to add mode tomodeName(required): Name for the new mode
- Updated collection with new mode ID
figma_rename_mode
Rename an existing mode in a collection.
When to Use:
- Fixing mode names
- Updating naming conventions
- Making mode names more descriptive
collectionId(required): Collection containing the modemodeId(required): Mode ID to renamenewName(required): New name for the mode
figma_batch_create_variables
Create multiple variables in a single operation — up to 50x faster than calling figma_create_variable repeatedly.
When to Use:
- Creating multiple design tokens at once (e.g., a full color palette)
- Importing variables from an external source
- Any time you need to create more than 2-3 variables
collectionId(required): Collection ID to create all variables invariables(required): Array of 1-100 variable definitions, each with:name(required): Variable name (use/for grouping)resolvedType(required):"COLOR","FLOAT","STRING", or"BOOLEAN"description(optional): Variable descriptionvaluesByMode(optional): Object mapping mode IDs to values
figma_batch_update_variables
Update multiple variable values in a single operation — up to 50x faster than calling figma_update_variable repeatedly.
When to Use:
- Updating many token values at once (e.g., theme refresh)
- Syncing variable values from an external source
- Any time you need to update more than 2-3 variables
updates(required): Array of 1-100 updates, each with:variableId(required): Variable ID to updatemodeId(required): Mode ID to update value invalue(required): New value (COLOR: hex"#FF0000", FLOAT: number, STRING: text, BOOLEAN: true/false)
figma_setup_design_tokens
Create a complete design token structure in one atomic operation: collection, modes, and all variables.
When to Use:
- Setting up a new design system from scratch
- Importing CSS custom properties or design tokens into Figma
- Creating themed token sets (Light/Dark) with all values at once
- Bootstrapping a new project with a full token foundation
collectionName(required): Name for the new collectionmodes(required): Array of 1-4 mode names (first becomes default)tokens(required): Array of 1-100 token definitions, each with:name(required): Token name (use/for grouping)resolvedType(required):"COLOR","FLOAT","STRING", or"BOOLEAN"description(optional): Token descriptionvalues(required): Object mapping mode names (not IDs) to values. A value can be a literal or a DTCG brace reference ("{color.blue.600}", set-qualified forms like"{primitives.color.blue.600}"too) — resolved viacreateVariableAliasagainst both variables created in the same call and variables that already exist in the file.
tokens array). Unresolvable references warn per-item without failing the batch. Semantic collections no longer require raw figma_execute scripting.
Returns:
"Light", "Dark") instead of mode ID — the tool resolves names to IDs internally.
Performance: Creates everything in a single Plugin API roundtrip. Ideal for bootstrapping entire token systems.
🧩 Component Tools
⚠️ Requires Desktop Bridge Plugin: These tools require the Desktop Bridge plugin running in Figma. In Local Mode, the plugin connects via WebSocket. In Cloud Mode, pair first using figma_pair_plugin to connect through the cloud relay.
figma_search_components
Search for components by name or description. Supports both local file search and cross-file published library search.
When to Use:
- Finding existing components to instantiate
- Discovering available UI building blocks
- Searching a published design system library from another file
- Checking if a component already exists before creating
query(optional): Search term to match against component names or descriptionscategory(optional): Filter by categorylibraryFileKey(optional): File key of a published library for cross-file searchlibraryFileUrl(optional): URL of a published library file (alternative to libraryFileKey)limit(optional): Max results (default: 10, max: 25)offset(optional): Pagination offset
- Array of matching components with keys, names, variant info, and
source(“local” or “library”)
FIGMA_ACCESS_TOKEN environment variable.
figma_get_library_components
Discover published components from a shared/team library file. This is the primary tool for cross-file design system workflows.
When to Use:
- Browsing all components in a published design system
- Getting component keys for instantiation from another file
- Auditing a library’s component inventory with variant detail
libraryFileUrl(optional): URL of the library filelibraryFileKey(optional): File key of the library filequery(optional): Filter by component name or descriptionlimit(optional): Max results (default: 25, max: 100)offset(optional): Pagination offsetincludeVariants(optional): Include individual variant components (default: false)
- Component sets with variant counts and keys, standalone components, summary stats, and instantiation examples
- Call
figma_get_library_componentswith your design system file - Find the component you want and note its
key - Call
figma_instantiate_componentwith thatcomponentKey— the component is imported from the published library automatically
FIGMA_ACCESS_TOKEN environment variable. Local mode only.
figma_get_library_component_by_key
Resolve a single library component to its full property definitions, variants, and visual specs — using only the component key, with no need to first find the source library file’s URL.
This is the missing link between “I see a component key in search results” (from figma_search_components, figma_get_library_components, or the official Figma MCP’s search_design_system) and “I’m ready to instantiate a specific variant.”
When to Use:
- You have a component key (40-char hex) from any search tool and want to inspect what properties / variants it exposes before instantiating
- You need each variant’s published key so
figma_instantiate_componentcan target a specific one - You want per-variant visual specs (fills, strokes, padding, typography) for code-generation fidelity
- Tries Figma REST
/v1/component_sets/{key}first (most common case — buttons, inputs, anything with variants) - On 404, falls back to
/v1/components/{key}(standalone components) - Extracts
file_key+node_idfrom the response - Fetches the node at
depth=2to readcomponentPropertyDefinitionsand the variant subtree - For COMPONENT_SETs, also fetches the source file’s
/componentslist in parallel so each variant child node can be mapped to its published variant key
componentKey(required): The 40-char hex component key from search results. Works for both COMPONENT_SET and standalone COMPONENT keys.includeVisualSpecs(optional, defaulttrue): Include per-variant fills/strokes/padding/typography. Auto-stripped if the response would exceed 500KB.format(optional,"full"|"summary", default"full"):summaryomits per-variant visual specs. Auto-downgrades on large responses.
resolvedAs:"COMPONENT_SET"or"COMPONENT"fileKey+nodeId: source file + node idname,description,thumbnail_url,containing_frame,user,created_at,updated_atproperties:componentPropertyDefinitions(VARIANT / BOOLEAN / TEXT / INSTANCE_SWAP)variants[]: each withname,nodeId,key(the value to pass tofigma_instantiate_component), and optionalvisualSpecvisualSpec: root-level fills/strokes/effects/padding/typographybounds: width × height of the component setcompression(only when stripped):{ originalSizeKB, finalSizeKB, strippedVisualSpecs: true }warnings[]: non-fatal issues (e.g. variant-key resolution skipped)
FIGMA_ACCESS_TOKEN with library_assets:read + files:read scopes.
figma_get_library_variables
List every variable from team libraries the current file has subscribed. Uses the Plugin API path (figma.teamLibrary.getAvailableLibraryVariableCollectionsAsync() + getVariablesInLibraryCollectionAsync()), which works on every Figma plan — unlike the REST /v1/files/{key}/variables/local endpoint which is Enterprise-only.
When to Use:
- Inventory what design tokens (colors, spacing, typography sizes) are available from your subscribed libraries
- Find the
keyof a specific library variable so you can import it - Build a design system overview that includes shared tokens (not just file-local ones)
libraryName(optional): Case-insensitive substring filter on the source library’s namecollectionName(optional): Case-insensitive substring filter on the collection name within a libraryresolvedType(optional,"COLOR" | "FLOAT" | "STRING" | "BOOLEAN"): Filter to a single token type. Auto-prunes empty collections after filtering.
summary:{ totalCollections, totalVariables }filters: echoed back so you can verifycollections[]:[{ libraryName, collectionKey, collectionName, variableCount, variables: [{ key, name, resolvedType }] }]
figma_import_library_variable
Import a single variable from a subscribed library into the current file. After import, the variable becomes locally addressable by its returned id and can be passed to any tool that binds variables to nodes (figma_set_fills, figma_update_variable, etc.). Idempotent — calling twice returns the same local id.
When to Use:
- After
figma_get_library_variables, you’ve picked a token and want to use it in the current file - You’re building a composition that mixes file-local variables with library tokens
variableKey(required): The variable’s library key fromfigma_get_library_variables(thecollections[].variables[].keyfield — distinct from the localidyou’ll get back).
imported:{ id, key, name, resolvedType, description, variableCollectionId, remote }usage.bind: a hint string showing how to use the returnedidwith binding tools
- If the source library isn’t subscribed by the current file, returns a specific hint pointing to Figma > Assets panel > Libraries. The Plugin API rejects with a generic message; this tool detects the pattern and explains it.
figma_get_component_details
Get detailed information about a specific component.
Usage:
componentKey(required): The component’s key identifier
- Full component details including properties, variants, and metadata
figma_instantiate_component
Create an instance of a component on the canvas.
When to Use:
- Adding existing components to your design
- Building compositions from component library
- Creating layouts using design system components
componentKey(required): Component key to instantiatex(optional): X position on canvasy(optional): Y position on canvasoverrides(optional): Property overrides for the instance
- Created instance with node ID
figma_create_component_set
Create a component set with variants in one declarative call — replaces hand-written figma_execute + figma.combineAsVariants() scripts.
When to Use:
- Building a component’s full variant matrix (states × sizes × …) from a single base component
- Combining existing loose components into a proper component set
- Any time you’d otherwise script
combineAsVariantsby hand
- Generate from a base component — pass
baseComponentId+properties(the variant axes matrix). The base is cloned for every combination of the axes, each variant namedProp=Valuecomma-joined (e.g.State=hover, Size=sm), then combined into a set. The base component itself becomes the first variant (same node ID), so existing instances of the base survive as instances of that variant. - Combine existing components — pass
componentIds, optionally withvariantProperties(aligned 1:1 by index) to rename each component toProp=Valueform before combining. Property names and values must not contain=or,. On any pre-combine failure, a unified rollback restores the components’ original names.
baseComponentId(optional): Node ID of an existing COMPONENT to use as the base. Mutually exclusive withcomponentIds.properties(required withbaseComponentId): Variant property axes —{ State: ["default", "hover"], Size: ["sm", "lg"] }. Max 100 combinations.componentIds(optional): Node IDs of existing COMPONENT nodes to combine. Components already inside a component set are rejected.variantProperties(optional, only withcomponentIds): One property map per component, aligned by index. Without it, existing names are kept (names lacking=becomeProperty 1=<name>).name(optional): Name for the component set (defaults to Figma’s derived name)parentId(optional): Container frame/section to create the set in (defaults to current page)position(optional):{ x, y }position within the parentautoArrange(optional, defaultfalse): Lay the new set out as a labeled grid (columns = last property, rows = other properties) inside a white container — same in-place layout asfigma_arrange_component_setarrangeOptions(optional):gap,cellPadding,columnProperty— used whenautoArrangeis true
Prop=Value names; they live on the SET (componentPropertyDefinitions), not on individual variants.
Size guidance: hard cap 100 variants. The timeout auto-scales with variant count (~1.2s/variant, 30s floor / 2min cap) at every hop, but above ~40 variants the single-pass clone+combine gets slow and the response carries a warning — prefer splitting large matrices into multiple sets (e.g. one set per Size).
Returns:
- The component set’s
id,name, and key - Each variant’s
name,nodeId, andkey— instantiate with a variant’s key viafigma_instantiate_component, not the set’s key
figma_arrange_component_set
Organize component variants into a professional component set with labels and proper structure.
When to Use:
- After creating multiple component variants
- Organizing messy component sets
- Adding row/column labels to variant grids
- Getting the purple dashed border Figma styling
componentSetId(optional): ID of component set to arrange (uses selection if not provided)componentSetName(optional): Find component set by nameoptions(optional): Layout optionsgap: Gap between grid cells (default: 24)cellPadding: Padding inside each cell (default: 20)columnProperty: Property to use for columns (default: auto-detect, usually “State”)
- Arranged component set with:
- White container frame with title
- Row labels (vertically centered)
- Column headers (horizontally centered)
- Purple dashed border (Figma’s native styling)
figma_create_slot
Add a Figma Slot to a component via the GA createSlot() API. The linked SLOT component property is created automatically and named after the slot — renaming the slot later renames the property.
Mode: Local / Cloud (requires Desktop Bridge)
figma_get_slots
List slots on a COMPONENT, COMPONENT_SET (aggregated across variants, each tagged with variantId/variantName), or INSTANCE. Returns ids, names, property keys, dimensions, layout mode, and current children. Use on an instance before figma_append_to_slot to discover slot names.
Mode: Local / Cloud (requires Desktop Bridge)
figma_append_to_slot
Populate a slot on a component instance. Slot content cannot be set through figma_set_instance_properties — Figma rejects slot values there by design; this tool is the population path.
Mode: Local / Cloud (requires Desktop Bridge)
clearExisting only clears after the new content validates.
figma_reset_slot
Clear all content from a slot on an instance. Takes slotId or instanceId + slotName.
Mode: Local / Cloud (requires Desktop Bridge)
figma_add_slot_property
Retrofit an existing frame as a slot: adds a SLOT component property and binds the frame to it via componentPropertyReferences.slotContentId. Prefer figma_create_slot for new slots. Supports description and preferredValues (the components the slot should accept). Works on standalone components and on component sets (bind a frame inside any variant). Existing property references on the frame are preserved.
Mode: Local / Cloud (requires Desktop Bridge)
figma_set_description
Add or update a description on a component, component set, or style.
When to Use:
- Documenting components for developers
- Adding usage guidelines
- Writing design system documentation
nodeId(required): Node ID of component/style to documentdescription(required): Description text (supports markdown)
- Confirmation with updated node info
🔧 Node Manipulation Tools
figma_resize_node
Resize a node to specific dimensions.
Usage:
figma_move_node
Move a node to a specific position.
Usage:
figma_clone_node
Create a copy of a node.
Usage:
- New node ID of the clone
figma_delete_node
Delete a node from the canvas.
Usage:
figma_rename_node
Rename a node.
Usage:
figma_set_text
Set the text content of a text node.
Usage:
figma_set_fills
Set the fill colors of a node.
Usage:
figma_set_strokes
Set the stroke colors of a node.
Usage:
figma_create_child
Create a child node inside a parent.
Usage:
🏷️ Component Property Tools
figma_add_component_property
Add a new property to a component.
Usage:
nodeId(required): Component node IDpropertyName(required): Name for the new propertypropertyType(required):"BOOLEAN","TEXT","INSTANCE_SWAP", or"VARIANT"defaultValue(required): Default value for the property
figma_edit_component_property
Edit an existing component property.
Usage:
figma_delete_component_property
Remove a property from a component.
Usage:
📦 Design System Kit
figma_get_design_system_kit
Extract your entire design system — tokens, components, and styles — in a single call. This is the preferred tool for design system extraction, replacing separate calls to figma_get_variables, figma_get_component, and figma_get_styles.
Returns component visual specs (exact colors, padding, typography, layout), rendered screenshots, token values per mode (light/dark), and resolved style values. Ideal for AI code generation — the visualSpec data provides pixel-accurate reproduction data.
Available in both Local and Remote modes.
figma_audit_design_system_report
Run a deterministic, Lighthouse-style health audit of the current design system and get the scored report back as data — no UI, no MCP Apps support, no ENABLE_MCP_APPS flag required. Same scoring engine as the Design System Dashboard app: Naming & Semantics, Token Architecture, Component Metadata, Accessibility, Consistency, Coverage — each 0–100, weighted into an overall score.
- Component data is fetched live-first through the Desktop Bridge — one page per plugin roundtrip (30s cap each, failures isolated per page) — with the REST published-library endpoints as fallback. The chosen source is disclosed in the report (
bridge-live/rest-published/none) because a published snapshot can be stale: scores from different sources are not comparable. - The bridge crawl is fileKey-verified: if the plugin is connected to a different file than requested, the audit refuses the data and falls back to REST rather than silently scoring the wrong file.
- Raw audit data caches for 5 minutes, so a summary call plus per-category drill-downs cost one crawl.
- The default summary output stays bounded regardless of file size; use
categoryfor detail instead offormat: "full"when working interactively.
design: auto-fixable via write tools like figma_rename_variable / figma_set_description / figma_rename_mode), can fix it after a design decision (design-assisted: e.g. contrast hues, alias tiering), or the work needs a human (manual: e.g. missing core components). Reports name the exact tools, so the natural next step is asking your agent to apply the fixes.
Local Mode only (requires the Desktop Bridge for live data; REST fallback works with a Figma token).
Usage:
Format options:
full— Complete data with visual specs and resolved style values. Best for implementing specific components.summary— Strips variant-level visual specs (medium payload). Good for overview + a few deep-dives.compact— Only names, types, and property definitions. Best for large design systems or getting an inventory.
tokens— Variables grouped by collection, with full mode support (light/dark/etc.)components— Published components with property definitions, variant specs, and visual specs (fills, strokes, effects, corner radius, layout, typography)styles— Color, text, and effect styles with resolved valuesai_instruction— Guidance for the AI on how to use the extracted dataerrors— Any sections that failed to extract (partial results are still returned)
📊 Design System Summary Tools
figma_get_design_system_summary
Get a high-level overview of the design system in the current file.
Usage:
- Component count and categories
- Variable collections and counts
- Style summary (colors, text, effects)
- Page structure overview
figma_get_token_values
Get all variable values organized by collection and mode.
Usage:
- Variables organized by collection
- Values for each mode
- Variable metadata
AI Decision Guide: Which Tool to Use?
For Design System Extraction
Tip: Preferfigma_get_design_system_kitover callingfigma_get_variables,figma_get_component, andfigma_get_stylesseparately. It returns all three in a single optimized call with visual specs and resolved values.
For Design Creation
For Variable Management
For Design-Code Parity
Prerequisites Checklist
Before using write tools, ensure one of the following: Local Mode:- ✅ Running in Local Mode (NPX/Git)
- ✅ Desktop Bridge plugin is running in your Figma file
- ✅
figma_get_statusreturnssetup.valid: true
- ✅ Desktop Bridge plugin is running in your Figma file with Cloud Mode enabled
- ✅ Paired via
figma_pair_plugin(or natural language: “connect to my Figma plugin”)
🔍 Design-Code Parity Tools
figma_check_design_parity
Compare a Figma component’s design specs against your code implementation. Produces a scored parity report with actionable fix items.
When to Use:
- Before sign-off on a component implementation
- During design system audits to catch drift between design and code
- To verify that code accurately reflects the design spec
fileUrl(optional): Figma file URL (uses current URL if omitted)nodeId(required): Component node IDcodeSpec(required): Structured code-side data with sections:visual: backgroundColor, borderColor, borderRadius, opacity, shadow, etc.spacing: paddingTop/Right/Bottom/Left, gap, width, height, minWidth, maxWidthtypography: fontFamily, fontSize, fontWeight, lineHeight, letterSpacing, colortokens: usedTokens array, hardcodedValues array, tokenCoverage percentagecomponentAPI: props array (name, type, required, defaultValue, description)accessibility: role, ariaLabel, keyboardInteraction, focusManagement, contrastRatiometadata: name, filePath, version, status, tags, description
canonicalSource(optional): Which source is truth —"design"(default) or"code"enrich(optional): Enable token/enrichment analysis (default: true)
summary: Total discrepancies, parity score (0-100), counts by severity (critical/major/minor/info), categories breakdowndiscrepancies: Array of property mismatches with category, severity, design value, code value, and suggestionactionItems: Structured fix instructions specifying which side to fix, which Figma tool or code change to applydesignData: Raw Figma data extracted from the component (fills, strokes, spacing, properties)codeData: The codeSpec as providedai_instruction: Structured presentation guide for consistent report formatting
score = max(0, 100 - (critical×15 + major×8 + minor×3 + info×1))
COMPONENT_SET Handling:
When given a COMPONENT_SET node, the tool automatically resolves to the default variant (first child) for visual comparisons (fills, strokes, spacing, typography). Component property definitions and naming are read from the COMPONENT_SET itself.
figma_generate_component_doc
Generate platform-agnostic markdown documentation for a component by merging Figma design data with code-side info. Output is compatible with Docusaurus, Mintlify, ZeroHeight, Knapsack, Supernova, and any markdown-based docs platform.
When to Use:
- Generating design system component documentation
- Creating developer handoff documentation
- Building a component reference library
fileUrl(optional): Figma file URL (uses current URL if omitted)nodeId(required): Component node IDcodeInfo(optional): Code-side documentation info. Read the component source code first, then fill in relevant sections:importStatement: Import pathfilePath: Component file pathpackageName: Package nameprops: Array of prop definitions (name, type, required, defaultValue, description)events: Array of event definitions (name, payload, description)slots: Array of slot/sub-component definitions (name, description)usageExamples: Array of code examples (title, code, language)changelog: Version history entries (version, date, changes)variantDefinition: CVA or variant definition code block (rendered in Implementation section)subComponents: Array of composable sub-parts (name, description, element, dataSlot, props)sourceFiles: Array of related files (path, role, variants, description) — used for Source Files table and Storybook link detectionbaseComponent: Base component attribution (name, url, description) — e.g., “Built on shadcn/ui Alert”
sections(optional): Toggle individual sections on/off (overview, statesAndVariants, visualSpecs, implementation, accessibility, changelog)outputPath(optional): Suggested file path for savingsystemName(optional): Design system name for documentation headersenrich(optional): Enable enrichment analysis (default: true)includeFrontmatter(optional): Include YAML frontmatter metadata (default: true)history(optional): Pull an ongoing changelog instead of relying on hand-writtencodeInfo.changelogentries. Both sources are off by default, so existing callers are unaffected.figma(defaultfalse): Walk Figma version history and diff each consecutive pair scoped to this component, producing one row per version that actually changed itgit(defaultfalse): Rungit logfor the component’s source files. Local mode only — the Cloudflare Worker runtime has no filesystem or git binaryversions(default5, max20): How many Figma versions to walk backincludeAutosaves(defaultfalse): Include unlabeled Figma auto-saves. Prefers labeled versions, but auto-falls back to auto-saves when a file has none (see below), so you rarely need to set thismode(summary|standard|detailed, defaultstandard):detailednames individual component properties and variable bindings instead of counting themgitLimit(default10, max50): How many commits to listgitPaths(optional): Explicit paths to log. Defaults tocodeInfo.filePathplus everycodeInfo.sourceFiles[].pathrepoPath(optional): Repo directory to run git in. Defaults to the server’s working directory
componentName: Resolved component namemarkdown: Complete markdown documentation with frontmatter, overview, states & variants, visual specs, implementation, accessibility sectionsincludedSections: Which sections were generateddataSourceSummary: What data sources were available (Figma enriched, code info, variables, styles)historySummary: Present only whenhistorywas requested — per-source row counts, API calls made, resolved git paths, and any degradation notessuggestedOutputPath: Where to save the fileai_instruction: Guidance for the AI on next steps (saving file, asking user for path)
Component History (history)
When either history source is enabled, the generated doc gains a ## History section in place of the pass-through ## Changelog:
codeInfo.changelog entries you pass are still rendered, folded in as Release notes. Frontmatter also gains figmaVersion / figmaVersionDate provenance — kept separate from the code-side version semver.
Cost: design history costs roughly one API call per version walked (N rows needs N+1 scoped node snapshots). Past-version snapshots are immutable and cached per process, so repeat runs on the same component are nearly free.
Scoping: history tracks the COMPONENT_SET when the node belongs to one. Variant node IDs churn as variants are added and removed, so the set is the stable identity across versions.
Labeled versions vs auto-saves: labeled versions make the best changelog rows, but many real design-system files have none at all — a mature file was verified live with 72 auto-saves and 0 labeled versions. Rather than emit an empty section there, history falls back to auto-saves automatically and says so in a note (historySummary.design.usedAutosaveFallback). Auto-save noise is largely absorbed downstream: a version only becomes a row if it actually changed the scoped component — on a real 24-variant Button, 20 version-pairs produced 4 rows. Auto-save rows render as _(auto-save)_ with the date rather than the raw 19-digit version ID; that ID stays available as historySummary.design.latestVersionId.
Coverage limits — the generated doc states these inline so an empty table is never misread as “nothing changed”:
- Figma’s REST version snapshots omit description and Dev Mode annotation edits, raw layout/visual properties, and variable value changes. Structure (child layers), component property definitions, variable bindings, and renames are tracked, at depth 2
- Version-history retention is plan-dependent, so lower tiers expose a shorter window
- Reading version history requires the
file_versions:readOAuth scope, or the Versions Read permission on a Personal Access Token. Without it the section degrades to an explanatory note rather than failing the doc githistory uses--follow(renames traced) only when exactly one path is resolved — git supports it for a single pathspec only
💬 Comment Tools
figma_get_comments
Get comments on a Figma file. Returns comment threads with author, message, timestamps, and pinned node locations.
When to Use:
- Reviewing feedback threads on a design file
- Checking for open comments before a release
- Retrieving comment IDs to reply to or delete
fileUrl(optional): Figma file URL (uses current URL if omitted)as_md(optional): Return comment message bodies as markdown (default: false)include_resolved(optional): Include resolved comment threads (default: false)
comments: Array of comment objects withid,message,user,created_at,resolved_at,client_meta(pinned location)summary: Total, active, resolved, and returned counts
figma_post_comment
Post a comment on a Figma file, optionally pinned to a specific design node. Supports replies to existing threads.
When to Use:
- After
figma_check_design_parityto notify designers of drift - Leaving feedback on specific components or elements
- Replying to an existing comment thread
fileUrl(optional): Figma file URL (uses current URL if omitted)message(required): The comment message textnode_id(optional): Node ID to pin the comment to (e.g.,'695:313')x(optional): X offset for comment placement relative to the nodey(optional): Y offset for comment placement relative to the nodereply_to_comment_id(optional): ID of an existing comment to reply to
comment: Created comment object withid,message,created_at,user,client_meta
figma_delete_comment
Delete a comment from a Figma file by its comment ID.
When to Use:
- Cleaning up test or outdated comments
- Removing resolved feedback after fixes are confirmed
- Managing comment threads programmatically
fileUrl(optional): Figma file URL (uses current URL if omitted)comment_id(required): The ID of the comment to delete (get IDs fromfigma_get_comments)
success: Boolean indicating deletion successdeleted_comment_id: The ID that was deleted
📝 Annotation Tools
Annotation tools require the Desktop Bridge plugin to be running in Figma. Annotations are distinct from comments: they are node-level design specs that can pin specific properties (fills, width, typography, etc.) and support markdown-formatted labels. Designers use them to communicate animation timings, accessibility requirements, interaction specs, and other implementation details.figma_get_annotations
Read annotations from a Figma node. Annotations are designer-authored specs attached to nodes — they can include notes (plain text or markdown), pinned design properties (fills, width, fontSize, etc.), and category labels.
Mode: Local / Cloud
When to Use:
- Discovering designer specs on a component before implementation
- Reading animation timings, interaction behaviors, or accessibility requirements
- Getting all annotations across a component tree for documentation
Returns:
nodeId,nodeName,nodeType: The target node infoannotations: Array of annotations withlabel,labelMarkdown,properties(pinned design properties),categoryId,categoryNameannotationCount: Number of annotations on this nodechildren: (when include_children=true) Array of child nodes with their annotationschildAnnotationCount: Total annotations across childrenavailableCategories: List of annotation categories in the file
figma_set_annotations
Write or clear annotations on a Figma node. Supports plain text labels, rich markdown labels, pinned design properties, and annotation categories. This operation is undoable in Figma (Cmd+Z).
Mode: Local / Cloud
When to Use:
- Documenting animation timings and easing curves on components
- Adding accessibility requirements to design nodes
- Communicating implementation notes from design reviews
- Clearing outdated annotations after implementation is complete
Annotation object fields:
Returns:
success: Boolean indicating write successnodeId: The target node IDnodeName: The node nameannotationCount: Number of annotations after the operationmode: The write mode used
Note: Pinned properties must be valid for the node type. For example,cornerRadiusworks on COMPONENT nodes but not on COMPONENT_SET nodes. Usefigma_get_annotation_categoriesto discover valid category IDs.
figma_get_annotation_categories
List available annotation categories in the current Figma file. Categories group annotations by purpose (e.g., interactions, accessibility, development notes).
Mode: Local / Cloud
When to Use:
- Discovering available categories before creating annotations
- Listing category IDs for use with
figma_set_annotations
categories: Array of{ id, name }category objects
Annotations Workflow
🖼️ Image Tools
figma_set_image_fill
Set an image fill on one or more Figma nodes. Accepts base64-encoded image data or (in Local Mode) an absolute file path.
Mode: Local / Cloud
When to Use:
- Applying photos, illustrations, or textures to frames and shapes
- Setting hero images, avatars, or background images
- Replacing placeholder images with real assets
nodeIds(required): Array of node IDs to apply the image fill toimageData(required): Base64-encoded image data (JPEG/PNG), or an absolute file path starting with/(Local Mode only)scaleMode(optional): How the image fills the node —"FILL"(default),"FIT","CROP", or"TILE"
imageHash: Figma’s internal hash for the created imageupdatedCount: Number of nodes successfully updatednodes: Array of updated node IDs and names
🔍 Accessibility Tools
Three tools provide full-spectrum accessibility coverage across design and code — without maintaining a rule database. Design-side checks are bounded by Figma’s API; code-side checks delegate to axe-core (Deque).figma_lint_design
Run comprehensive WCAG 2.2 accessibility and design quality checks on the current page or a specific node tree. Returns categorized findings with severity levels.
Mode: Local / Cloud
When to Use:
- Checking designs for WCAG accessibility compliance (14 checks)
- Finding hardcoded colors that should use design tokens
- Detecting detached components, missing focus variants, color-only states
- Auditing heading hierarchy, reading order, reflow readiness
- Pre-handoff quality checks
nodeId(optional): Node ID to lint (defaults to current page)rules(optional): Rule filter —["all"](default),["wcag"](14 rules),["design-system"],["layout"], or specific rule IDsmaxDepth(optional): Maximum tree depth to traverse (default: 10)maxFindings(optional): Maximum findings before stopping (default: 100)
Each finding includes a
wcagLevel field (a, aa, or best-practice) so teams can filter by their target conformance level.
Individual Rules:
Returns:
- “Check my design for accessibility issues”
- “Lint this page”
- “Find hardcoded colors”
- “Are there any detached components?”
- “Run a WCAG contrast check”
- “Audit the design quality”
figma_audit_component_accessibility
Deep accessibility audit for a specific component or component set. Produces a scorecard covering state coverage, focus indicator quality, non-color differentiation, target size consistency, annotation completeness, and color-blind simulation.
Mode: Local / Cloud
When to Use:
- Validating a component’s accessibility before design handoff
- Checking if all interactive states (focus, disabled, error) are present
- Verifying color-blind safety with protanopia/deuteranopia/tritanopia simulation
- Auditing whether components have accessibility documentation
nodeId(optional): Node ID of a COMPONENT_SET, COMPONENT, or INSTANCE. Falls back to current selection.targetSize(optional): Minimum touch target size in px (default: 24 per WCAG 2.5.8). Use 44 for iOS, 48 for Android.
figma_scan_code_accessibility
Scan HTML code for accessibility violations using axe-core (Deque). Runs structural/semantic checks via JSDOM — no browser needed. Visual rules (color contrast) are disabled since they’re handled by figma_lint_design.
Mode: Local / Cloud (standalone — no Figma connection required)
When to Use:
- Scanning component HTML for ARIA, label, and semantic issues
- Checking code accessibility before merging
- Generating a CodeSpec for design-to-code parity comparison
- Validating that implemented code matches design accessibility intent
html(required): HTML string to scan (fragment or full document)tags(optional): WCAG tag filter —["wcag2a"],["wcag2aa"],["wcag22aa"],["best-practice"]context(optional): CSS selector to scope the scanmapToCodeSpec(optional): If true, auto-generatescodeSpecAccessibilityfor use withfigma_check_design_parityincludePassingRules(optional): Include pass/incomplete counts
📌 FigJam Tools
FigJam tools only work when the Desktop Bridge plugin is running in a FigJam board (editorType === 'figjam'). They return clear errors when used in Figma Design files.
figjam_create_sticky
Create a sticky note on a FigJam board.
Mode: Local / Cloud
Parameters:
figjam_create_stickies
Batch create multiple sticky notes (max 200). Font is loaded once for the entire batch.
Mode: Local / Cloud
Parameters:
figjam_create_connector
Connect two nodes with a connector line. Use node IDs from creation results.
Mode: Local / Cloud
Parameters:
figjam_create_shape_with_text
Create a labeled shape for flowcharts and diagrams.
Mode: Local / Cloud
Parameters:
figjam_create_table
Create a table with optional cell data.
Mode: Local / Cloud
Parameters:
figjam_create_code_block
Create a code block for sharing snippets and technical documentation.
Mode: Local / Cloud
Parameters:
figjam_auto_arrange
Arrange nodes in a grid, horizontal row, or vertical column layout.
Mode: Local / Cloud
Parameters:
figjam_get_board_contents
Read all content from a FigJam board. Returns stickies, shapes, connectors, tables, code blocks, and sections with their text content, positions, and type-specific properties (colors, shape types, cell data, connector endpoints).
Mode: Local / Cloud
Parameters:
Returns:
nodes— Array of node objects with id, type, name, position, dimensions, and type-specific datatotalFound— Number of nodes returnedtruncated— Whether results were capped at maxNodespage— Current page name
figjam_get_connections
Read the connection graph from a FigJam board. Returns all connectors as edges with their start/end node references and labels, plus a lookup of connected nodes.
Mode: Local / Cloud
Parameters: None
Returns:
edges— Array of{connectorId, startNodeId, endNodeId, label}connectedNodes— Map of node ID →{id, type, name, text}totalConnectors— Number of connectors foundtotalConnectedNodes— Number of unique connected nodes
☁️ Cloud Relay
figma_pair_plugin
Generate a pairing code to connect the Figma Desktop Bridge plugin to the cloud relay. This enables write operations from web-based AI clients.
Mode: Cloud only (available on /mcp endpoint)
Parameters: None
Returns:
code— 6-character alphanumeric pairing code (uppercase, no ambiguous characters)expiresIn— Expiry time (5 minutes)- Instructions for the user
- “Connect to my Figma plugin”
- “Pair with my design file”
- “Set up the cloud connection”
- “Link Figma to this chat”
- Generates a unique 6-character code stored in KV with 5-minute TTL
- User enters code in the Desktop Bridge plugin’s Cloud Mode section
- Plugin connects via WebSocket to the cloud relay Durable Object
- All subsequent write tool calls route through the relay to the plugin
🕒 Version History Tools
Six tools that turn a Figma file from a static snapshot into a queryable history. They compose: list versions → diff → generate changelog → blame specific changes back to the version that introduced them. All cache aware (past versions are immutable, so repeat queries on the same range cost zero new API calls). Required scope:file_versions:read (OAuth) or Versions (Read) on a Personal Access Token, in addition to the standard file_content:read.
figma_get_file_versions
List a file’s version history with author, label, description, and timestamp metadata. Auto-paginates up to max_versions. Defaults to labeled-only (skips auto-saves) — pass include_autosaves: true to see every saved state.
Usage:
{ versions: [...], pagination: { has_more, next_cursor, returned, filtered_out_autosaves } }. Each version entry includes id, label, description, created_at, user.handle, and is_labeled.
figma_get_file_at_version
Snapshot a file (or selected nodes) as it existed at a past version_id. Thin wrapper over figma_get_file_data with the version param plumbed through.
Usage:
figma_diff_versions
Structured diff between two versions. Always returns a cheap page-structure diff (~2 API calls, parallel). When component_ids are passed, additionally produces per-node diffs at depth=2: added/removed children (variants), name/description changes, componentPropertyDefinitions changes, and boundVariables deltas.
Usage:
notes[].
figma_get_changes_since_version
Convenience wrapper for figma_diff_versions with to_version="current" (HEAD). Useful for “what’s changed since the last code-sync” workflows.
figma_generate_changelog
Markdown changelog generator. Wraps figma_diff_versions with author enrichment (one extra cheap API call to look up labels and authors for the from/to versions). Returns BOTH a markdown string ready for release notes / PRs / Storybook MDX, and the structured diff payload.
summary produces a one-line release note; standard includes sectioned page + per-component change counts; detailed includes per-property and per-binding bullets.
figma_blame_node
Find the version that introduced a specific change to a node — answers “who/when added this.” Walks history backward via binary search (~log₂(N) probes instead of N), so a 200-version lookback typically costs ~8 API calls instead of 200.
{ introduced_at: { version_id, label, created_at, user_handle, is_labeled }, attribution_certainty, summary, notes }.
attribution_certainty is one of:
"exact"— the introduction point is fully localized and authored by a real user"system_attributed"— the introducing version was a system-triggered autosave (user="Figma"); setinclude_autosaves: falseand re-run to find the labeled shipping author"exists_at_lookback_horizon"— the actual introduction is older thanmax_versions_to_walk; raise the cap and retry"metadata_unavailable"— introduction was atstart_versionitself but author lookup couldn’t reach it within the version-list lookback
notes[] on every response.
Error Handling
All tools return structured error responses:"FIGMA_ACCESS_TOKEN not configured"- Set up your token (see installation guide)"Failed to connect to browser"- Browser initializing or connection issue"Invalid Figma URL"- Check URL format"Node not found"- Verify node ID is correct"Desktop Bridge plugin not found"- Ensure plugin is running in Figma"Invalid hex color"- Check hex format (use #RGB, #RGBA, #RRGGBB, or #RRGGBBAA)