What is Figma Context MCP Cached?
Figma Context MCP Cached is an enhanced Figma Context MCP server that stores Figma file data on disk with a configurable time to live. Repeat requests read from the local cache instead of the Figma API, which eases rate limits, especially on free Figma accounts.
This server is an enhanced build of Figma Context MCP. It adds local persistent caching, which cuts the number of requests sent to the Figma API and so eases rate limiting while speeding up repeat reads.
The README targets free Figma accounts, frequent context requests and MCP clients such as Cursor. It calls itself a drop in replacement for the original, with no change to how a client calls it.
How the cache behaves
- The first request for a file pulls the full file from the Figma API and writes it to local disk.
- Until the TTL expires,
get_figma_dataanswers from the cache without calling Figma again. - After expiry the server refetches and refreshes the cache, or you force it with
forceRefresh: true.
The stale data risk
The README warns that edits made inside the TTL window may not appear. Page structure, component properties, added nodes and small copy changes can all return old data. Enable caching once a design is stable, not mid iteration.
Who it is for
Best for
- Free Figma accounts that keep hitting rate limits. Related: official Figma MCP and Framelink.
- Code generation and design review passes against finished files.
- Repeated reads of the same file across many agent sessions.
Not for
- Files in active redesign, where cached reads can hide the latest change.
- Anyone who wants to write to Figma. The listed tools only read data and download images.
- Teams that need no Figma token. An API key is still required.
Use it with UiChemy
A cached read speeds up a design to code loop. To turn the finished Figma design into a WordPress site, UiChemy's Figma to WordPress converts it into Elementor or Gutenberg pages. Click Get Started on uichemy.com.
Connect Figma Context MCP Cached: Disk Cache for Figma API Limits to your AI client
npx -y @pactortester/figma-mcp-cached --stdio --figma-api-key=YOUR_FIGMA_API_KEYCreate a Figma token
Create a Personal Access Token in your Figma developer settings and keep it out of version control.
Add the server to your MCP config
Paste the config above into your client's MCP settings. The README example uses Cursor and a 7 day TTL, and it switches caching on with the figma-caching argument.
Tune the cache if needed
Optional fields are cacheDir, autoCleanup, cleanupInterval, maxMemoryCacheSize and encryptionKey. Command line arguments win over environment variables.
Paste a Figma link
Give the agent a Figma URL. It calls
figma_prepare_filefirst, thenget_figma_data. Say refresh the file to force a fresh pull.
Client configuration
{
"mcpServers": {
"Figma-Context-MCP-Cached": {
"command": "npx",
"args": [
"-y",
"@pactortester/figma-mcp-cached",
"--stdio",
"--figma-api-key=YOUR_FIGMA_API_KEY",
"--figma-caching={\"ttl\":{\"value\":7,\"unit\":\"d\"}}"
]
}
}
}Endpoints and ports
~/Library/Caches/FigmaMcpLinux uses ~/.cache/figma-mcphttps://github.com/Pactortester/figma-mcp-cached/blob/main/CHANGELOG.mdRelease historyWhat you can do with it
- Persistent disk cacheFigma file data is saved locally with a TTL you set in seconds, minutes, hours or days.
- Prepare then fetchfigma_prepare_file readies the file and checks the node exists before get_figma_data reads it.
- Forced refreshSet forceRefresh to true, or tell the agent to pull the latest, to bypass the cache.
- Memory layerAn LRU memory cache and parsed node caching avoid repeat disk reads.
- Automatic cleanupExpired cache files are removed on a schedule, and cleanup_cache does it on demand.
- Optional encryptionA key enables AES-256-CBC encryption of cache files for sensitive designs.
Tools it exposes
figma_prepare_file
Checks the cache for a Figma URL, validates an optional nodeId and refetches when the cache is stale. Accepts forceRefresh.
get_figma_data
Returns metadata, the node tree, global style variables, components and component sets for a file or node.
download_figma_images
Downloads SVG and PNG assets from a file, to the system Downloads folder by default, with a configurable PNG scale.
list_cache
Shows whether caching is enabled, the cache directory, file count, total size and TTL.
cleanup_cache
Deletes expired and corrupted cache files and reports the size before and after.
Frequently asked questions
What is Figma Context MCP Cached?
It is an enhanced version of Figma Context MCP that saves Figma file data to local disk. Repeat requests read the cache instead of the Figma API, which reduces rate limit pressure and speeds up responses for MCP clients.
How do I install Figma Context MCP Cached?
Add the server to your MCP config with npx, the package name @pactortester/figma-mcp-cached, the stdio flag, your Figma API key and a figma-caching argument that sets the TTL. The README shows a Cursor example with a 7 day TTL.
Does it need a Figma API key?
Yes. The quick start tells you to create a Personal Access Token in Figma's developer settings and pass it with the figma-api-key argument. The cache reduces how often the API is called but does not replace the key.
Can the cache return outdated designs?
Yes. Within the TTL the server may return old data after you edit a page or component. The README advises enabling the cache once a design is stable, and using forceRefresh or deleting the cache file to pull the latest.
How do I force a refresh?
Call figma_prepare_file with forceRefresh set to true, or ask the agent for the latest design. You can also delete the matching cache file in the cache directory by hand. The next read then fetches from Figma again.
Which MCP clients does it work with?
The README example targets Cursor and describes the server as suited to Cursor and other MCP clients. It calls itself a drop in replacement for the original Figma Context MCP, with no change needed to client calls.
How does it differ from other Figma MCP servers?
It stays a Figma Context MCP server and adds disk caching on top of the original. It keeps the original's call pattern and adds a persistent local disk cache to cut API requests and ease rate limits. Related: Figma MCP Rust.