sqid MCP server
OfficialServes as the OAuth authorization server for the MCP server: clients discover it via protected-resource metadata, register as public clients, and obtain audience-scoped PKCE tokens. The server verifies those tokens and uses a Clerk machine (sqid-mcp) to mint short-lived machine tokens whose claims identify the signed-in person, so upstream Sqid console calls act as that user.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
sqid MCP server
Sqid for AI editors and agents over the Model Context Protocol: the same actions as
the sqid CLI, served as tools. Hosted, Streamable
HTTP, OAuth through Sqid's own sign-in. No API keys to paste.
claude mcp add --transport http sqid https://mcp.sqid.dev/mcpThe first call opens a browser to sign in to Sqid and approve the connection. After that the agent can create a site, get the snippet, wire webhooks, list identified visitors, enrich a contact, read moments and billing, as the person who approved it.
Tools
Tool | Does |
| Who approved the connection, role, plan present |
| Sites, the install snippet, the "stream is live" check |
| Where identified visitors (or moments, pages, clicks) are delivered |
| Identified visitors on demand |
| Email, phone or LinkedIn URL to a full record |
| Signals state, on or off, recent moments |
| Plan, card, balance, this month |
Card entry and plan changes are not tools. They happen in the console or the CLI.
Every tool carries MCP annotations: reads are readOnlyHint: true, the five that change
things (site_create, webhook_add, webhook_test, enrich, signals_set) are not, and
nothing is marked destructive because no tool deletes. Clients such as Claude Code use
these to decide what needs a confirmation.
Related MCP server: OutcomeCI MCP Server
How auth works
Clerk (the instance behind app.sqid.dev) is the OAuth authorization server. This server is
a resource server at MCP_RESOURCE_URL, per the MCP authorization spec: clients discover it
through /.well-known/oauth-protected-resource, register as public clients (Client ID
Metadata Documents, with dynamic registration as fallback), obtain a token whose audience
is that URL with PKCE, and send it on every POST /mcp. A missing or bad token gets a 401
with a WWW-Authenticate challenge pointing at the metadata.
The client's token is verified here and goes no further. MCP forbids passing it upstream,
so for calls to app.sqid.dev the server is a Clerk machine (sqid-mcp) and mints its own
short-lived machine token whose claims name the verified person. The console admits that
machine and acts as that person, logged as such. Nothing is stored on this server beyond
those tokens in memory.
Run
It is a Cloudflare Worker (src/worker.js, wrangler.jsonc), served at mcp.sqid.dev.
The same app also runs on Node (src/server.js) for local runs and tests.
npm install
npm test # offline: fake Sqid API, real MCP client
cp .env.example .dev.vars # local secrets for wrangler dev (never committed)
npm run dev # wrangler dev on http://localhost:8787
npm run check # wrangler deploy --dry-run
npm run deploy # wrangler deploySecrets are set once per Worker and never live in the repo:
wrangler secret put CLERK_PUBLISHABLE_KEY
wrangler secret put CLERK_SECRET_KEY
wrangler secret put MCP_MACHINE_SECRET_KEYHealth: GET /alive. Logs: wrangler tail.
Registering with Clerk
Clerk dashboard, Agents, MCP servers: add
MCP_RESOURCE_URLand allow public clients to connect. Clerk then issues tokens with that audience and shows the consent screen.Clerk dashboard, Configure, Machines: create a machine
sqid-mcp; its secret key isMCP_MACHINE_SECRET_KEY; its id goes into the console'sclerk.mcpMachineIds.
MIT
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
OAuth 2.1 short-link tools for AI agents with scoped tokens, approvals, audit logs, and revocation.
Real-time web analytics for AI agents: query traffic, funnels, revenue, and manage your sites.
Privacy-first, cookie-free web analytics. Read stats and manage sites from your AI.
Create, change and install StellaFrame website widgets from an AI agent. OAuth sign-in, no API key.
Related MCP Servers
- AlicenseAqualityDmaintenancePrivacy friendly, cookieless web analytics built MCP-first. "Add analytics to my Next.js app" → an AI agent runs the setup_analytics_for_site tool, picks the right install snippet, edits your layout file, and verifies the script is loading. OAuth onboarding, no API keys to paste.2851 npm1MIT

OutcomeCI MCP Serverofficial
AlicenseNot gradedqualityCmaintenanceEnables AI assistants to create, validate, and manage agent-powered workflows across team tools, with tools for inspecting schemas, configuring triggers, and reviewing runs and logs. It provides authorized remote access through OAuth without local installation or shared API keys.1MIT- AlicenseNot gradedqualityAmaintenanceEnables AI agents to interact with Webflow's Data API and Designer, letting them manage CMS content, pages, and site design elements through OAuth-authenticated remote or locally hosted connections.501 npmMIT
- AlicenseNot gradedqualityAmaintenanceEnables AI agents to interact with Webflow sites through the Webflow Data API, supporting content analysis, page and CMS updates, and Designer canvas changes via OAuth-authenticated remote or local setups.501 npmMIT