How to connect Figma to Hermes Agent

Pairing is a two-step handshake: run hermes clawlink begin, approve the link in your browser, then run hermes clawlink finish. If you ran finish before approving, or the approval link expired, run hermes clawlink begin again to get a fresh link. Confirm the plugin was installed with --enable, then verify with hermes clawlink test.

Reconnect Figma from the dashboard, then start a fresh chat if the runtime still has the old tool catalog loaded.

Figma tool schemas load on demand the first time a tool runs and are cached after that, so this error usually clears on its own: wait a few seconds and retry the same request. If every Figma call keeps failing with it in a fresh chat, reconnect from the dashboard, and contact support if it still persists — that pattern points to a configuration problem on our side, not something you can fix by reconnecting again.

Two usual causes. The connected account may not have access to the specific workspace, inbox, store, or project in the request — check that first. If access looks right, the agent may have sent a placeholder value (like "YOUR_ID" or an example id from documentation) instead of a real one: ask it to run a list or search tool first, then retry the action with a real id from those results. Most failures at this stage are one of these two, not ClawLink bugs.

Search results will tell you to check your X-Figma-Token header, regenerate an expired personal access token, or confirm a scope name like file_content:read. None of that applies here, because there is no token and the scopes were fixed at consent time. What causes it on a hosted connection is simpler. Most often the file was never shared with the account you connected, so the agent reads everything else you own and fails on that one, which looks exactly like a bug. Open the file's Share settings and confirm the connected account is listed. The second cause is a guessed id: file keys come out of the URL after figma.com/file/ or figma.com/design/, and an agent that invents one gets a 403 or 404 on a healthy connection. Run figma_discover_figma_resources with the Figma link and use the key it returns. A third and rarer cause is an Enterprise-only endpoint, covered in the tool-not-found card below.

Ask the agent to diagnose it:

Run figma_get_current_user and tell me which Figma account is connected. Then run figma_discover_figma_resources with the URL I paste and show me the file key it returns before retrying anything.

Figma rate-limits its REST API per user, per plan, per app, and the tier that catches people is the read cap on Starter-plan files: Figma's documentation puts Tier 1 read endpoints, meaning get file, get file nodes, and get image, at 6 calls a month for View and Collab seats, while Dev and Full seats get roughly 10 to 20 requests a minute instead. That is Figma's budget rather than a ClawLink limit, and a personal access token against the same file would spend the same allowance. The response carries a Retry-After header, so waiting is the correct response. If you are on a paid seat, the practical fix is to stop pulling whole files: pass ids and depth to figma_get_file_json so you fetch one subtree at a time.

Ask the agent to diagnose it:

Report the exact status code and body of the last failed Figma call, including any Retry-After value. Do not retry it.

If the tool name looks ordinary, the catalog had not loaded when the first call went out; schemas load on demand and retrying once usually resolves it. If the agent is asking specifically for variables tools, those are absent on purpose. Figma marks the file_variables read and write scopes as Enterprise-plan-only, so tools depending on them were removed rather than shipped in a state that fails for everyone below Enterprise. Design tokens still work without that scope: figma_extract_design_tokens reads styles and node-derived values on any plan. Note that this is a different failure from the one most search results describe, where a local Figma MCP server is missing from an editor's config file. There is no local server in this setup.

Ask the agent to diagnose it:

List the Figma tools you actually have available and tell me which one extracts design tokens. Use that name instead.

Try reconnecting Figma and complete the consent flow in the same browser session. Partial OAuth approvals or switching accounts mid-flow can leave the connection incomplete.

Yes. ClawLink is the fastest way to connect Hermes to Figma: link your Figma account once in the browser and Hermes Agent can call the Figma API through 37 ready-made tools — no custom code or token handling.

Install the plugin with hermes plugins install ClawLink-HQ/hermes-plugin --enable, then pair once: run hermes clawlink begin, approve the link in your browser, and run hermes clawlink finish. Connect Figma in the dashboard and Hermes can use it from the next message — no config files or API keys to manage.

About two minutes. Sign in, click Connect next to Figma in the dashboard, authenticate, and Hermes Agent can use it from the next chat message.

The alternative to ClawLink is usually manual OAuth app setup plus your own token handling, permission troubleshooting, and tool plumbing for Hermes Agent. That is fine if you want to build and maintain the integration yourself. Most teams just want Figma working from chat.

No, and avoiding one is most of the point. A Figma personal access token acts on your account, in Figma's own words, as if it were the user who generated it: every file you can reach, no narrowing, and no expiry unless you set one. It also has to live somewhere, which in practice means a config file on whatever machine runs your agent. The ClawLink connection is an OAuth grant with 12 granular scopes covering file content and metadata, comments, projects and libraries, and webhooks. It shows up in your Figma settings under connected apps, and revoking it there ends the access immediately.

Figma's remote MCP server is a good product and it solves a different problem. It is a server you point an MCP client at, which suits editors like Cursor or Claude Code where the client is already an MCP host and you want design context inside the editor. This connection puts Figma tools directly into Hermes chat instead, with no server address to configure and nothing to keep running. The other practical difference is coverage: Figma's server is built around reading design context for code generation, while the tools here also post comments, create webhooks, extract design tokens, and render images.

Files that are shared with the account you connected, and nothing else. Figma's permission model is not overridden by OAuth scopes, so a file nobody shared with you stays invisible to the agent exactly as it is to you. Within that, the grant covers reading file content and metadata, reading and posting comments, reading projects and libraries, and creating webhooks. It does not include a canvas-write scope, so the agent cannot edit or create frames, and it does not include the variables or dev-resources scopes, which Figma marks as Enterprise-plan-only.

No. Dev Mode enablement and the local endpoint at 127.0.0.1:3845 belong to Figma's desktop MCP setup, which is what most "Figma MCP tool not found" results are about. This connection talks to the Figma REST API through ClawLink, so there is no desktop app requirement, no local port, and no client config to re-add when it stops working.

Pairing is a two-step handshake: run hermes clawlink begin, approve the link in your browser, then run hermes clawlink finish. If you ran finish before approving, or the approval link expired, run hermes clawlink begin again to get a fresh link. Confirm the plugin was installed with --enable, then verify with hermes clawlink test.