AI integrations

Shorten, track and report without leaving Claude.

Available nowPaid plansRemote server, nothing to install

Connect once and Claude can create links on your domain, look up how they perform and prepare campaign link sets, asking before anything destructive. Works in Claude Code today; the same server speaks to every MCP client.

See all 12 tools

The MCP server is included in every paid plan, Pro at $9 a month and Team at $29. You create the token in Settings and paste it into your client once.

Setup

Pick your client

Every path ends the same way: the client asks our server for its tools, and your token tells us who the links belong to.

Claude Code
One command in your terminal. Replace YOUR_TOKEN with a token from Settings.
claude mcp add --transport http snipzr https://mcp.snipzr.com/mcp --header "Authorization: Bearer YOUR_TOKEN"
Claude, desktop and web
Add a custom connector under Settings, then Connectors, with the server address. Where the connector setup cannot take a token header yet, use Claude Code; a one-click connect is on our list.

To confirm it worked: ask your assistant to list the Snipzr tools. 12 names starting with snipzr_ means you are connected.

Written for agents too

Let your assistant install it

This page is plain, linkable text and listed in our llms.txt, so an assistant with web access can read it and do the setup with you. Or paste this prompt:

Set up the Snipzr MCP server in my client. The server address is https://mcp.snipzr.com/mcp and it authenticates with an Authorization: Bearer header. I will create the token at https://www.snipzr.com/settings and paste it when you ask. Then list the tools so we can confirm it works.

Tools

12 tools, built on the public API

snipzr_create_link
Create one short link: destination, optional domain, back-half and title.
links:write
snipzr_create_links_batch
Create up to 100 links in one call, for campaign sets and migrations.
links:write
snipzr_list_links
List links with text, domain and campaign filters; paginated.
links:read
snipzr_get_link
Read one link by its back-half.
links:read
snipzr_update_link
Change a destination or title; edits are never metered.
links:write
snipzr_delete_link
Delete a link; the client asks for confirmation first.
links:write
snipzr_get_link_stats
Per-day clicks and scans with bots counted separately.
stats:read
snipzr_get_campaign_report
Clicks rolled up by campaign across every link.
stats:read
snipzr_get_usage
The plan, remaining link allowance and reset date.
usage:read
snipzr_list_domains
Your short domains and their status.
domains:read
snipzr_get_qr_code
A scannable PNG QR code for an existing link.
links:read
snipzr_build_utm_link
Compose a UTM link that follows your workspace conventions.
links:read

Every tool carries read-only or destructive annotations, so clients ask before a delete and never before a read. Error messages are written for agents to recover on their own: quota errors name the remaining allowance and validation errors name the field.

Permissions and data

What the server can and cannot do

  • It acts only within the scopes on your token; a read-only token cannot create, change or delete anything.
  • The token is yours: create it, limit it and revoke it in Settings at any moment, and every call it makes counts against your own plan limits.
  • The server never logs your destinations, and link analytics stay cookie-less, exactly as in the app.
  • Nothing runs on your machine: it is one hosted endpoint, so there is nothing to install or update.

FAQ

Common questions

Which plan do I need?
Any paid plan. Tokens are created in Settings on Pro and Team; the links an agent creates count toward the same monthly allowance as links you create by hand.
Is pasting a token into a client safe?
The token can be limited to exactly the scopes the client needs, is shown only once, and can be revoked in Settings at any time. Treat it like a password; if a machine or client is shared, give it its own token so you can revoke just that one.
Can an agent delete my links?
Only with a token that carries links:write, and the delete tool is marked destructive, so well-behaved clients ask you first. It also refuses to run unless the agent explicitly confirms.
My client is not listed. Does it work?
If it speaks MCP over HTTP, yes: use the generic config. Clients that only support stdio can bridge with npx mcp-remote.