Paste MCP & AI Tools - Infinite clipboard for Claude, Codex and other AI tools
by•
Privacy-first, lightning fast, searchable, and avaialbe across all devices. Save snippets, sync securely, and boost productivity with smart shortcuts and instant paste history.
Replies
Typinator 10
Great! Will it be accessible via Setapp as well?
Remention
Paste
@alice_ro Yes, Paste 6.6 is already available on @Setapp
Paste
Hey there,
Paste helps people save, search, and organize everything they copy across Mac and iOS. Today, we’re launching Paste MCP, a new way to bring your clipboard history into AI tools like Claude, Codex, and Cursor.
AI tools work best when they understand what you’re working on. But that context is often spread across links, notes, screenshots, snippets, files, and random ideas you copy during the day. Paste MCP brings that context into your AI workflow through a built in local MCP server on your Mac.
You can ask your AI tool to find something you copied earlier, use saved context in a draft, or create a pinboard without leaving your chat.
A few examples:
Check Paste for the meeting notes I copied earlier and turn them into a team update.
Find the links I saved about onboarding and summarize the key points.
Create a pinboard for this project and add the relevant items.
Paste MCP runs locally on your Mac. You choose which AI tools can connect, and you can remove access anytime.
Big thanks to @chrismessina for hunting Paste again and supporting us on this launch.
We’d love to hear what you try first, what you find useful, and what you think is missing.
Thanks for checking it out!
Migroot
@protsenkoalexandra Congrats on the launch! Can Claude save things back to Paste, or is it only able to search clipboard history?
Paste
@protsenkoalexandra @kate_prasniak you can save things, organize pinboards, search and do everything a human could do but using AI
GlowPulse
Сlipboard history as AI context, that's a brilliant reframe. been a paste user forever and never thought about it this way. the amount of stuff i copy daily and never look at again is wild in retrospect.
Brila
Sounds cool! Definetely gonna try it, congrats on launch!
Paste
@ikalimullin thanks!
A local MCP server that exposes clipboard history as retrievable context is the right read on where the bottleneck actually is. The clipboard is the richest unstructured context a person produces in a day, and most agents never see it. The piece I would want spelled out before I connect Claude or Cursor: the clipboard is also where passwords, API keys, and one time codes pass through. Does the MCP surface inherit the same ignore rules that keep secure apps out of Paste history, or is secret filtering a separate layer on the server? And when a tool queries, does it pull full items or only the ones I have pinned, so the model never sees the long tail by default?
@zimasilevuyo excellent question. This was my concern as well. I am concerned about my clipboard living anywhere but on my local machine. Any access to someone's clipboard history over any length of time would effectively expose them to a significant takeover or worse.
The pinboard example is super interesting. Being able to tell the AI to dynamically organize my random clipboard scraps into a structured pinboard sounds like a massive productivity boost. Definitely giving this a spin today..
Emma Intelligence
Congratulations! Does it work with windsurfing? P.S. I love your project, by the way, and I'm an old user.
Paste
@xanderiang Windsurf can also connect to Paste using the same MCP, just add as a custom AI tool in Paste Settings -> MCP -> Add AI Tool and follow the instructions
I copy so much stuff throughout the day and lose track of half of it. clipboard history as searchable context for AI tools is one of those things that sounds obvious but nobody else has done it
Congrats! This is one of those tiny workflow pieces that probably matters more than it looks.
Most of us using Claude/Codex end up re-pasting the same repo notes, commands, and half-written instructions all day. The interesting bit is whether Paste becomes a shared memory for the work or another drawer full of snippets.
How are you thinking about stale context — stuff that was useful yesterday but wrong today?