Clippy is a friendly popup GPT using local LLMs that is aware of your current context and pastes responses right in the text field where you are working, eliminating copy pasting. You can use it with voice commands or dictate, all with local models. And it can create Linear tickets or edit your Google Clendar with your voice.
I am coming from a non-technical background, from film production, and a bit more than a year ago I started building things that I like, that are useful for me, and that are fun.
I built Clippy because I wanted to have a cool way to dictate with local models and later I realised the whole GPT context copy and direct pasting possibility. And then ideas started storming with linear and co. I have a newer version with control of apple calendar and reminders, but thats still a work in progress.
And I chose Clippy for a vintage cute touch.
I did all the visuals myself in Figma and experimented a lot with liquid glass and local models.
Ah and as a fun fact - I coded around 90% of the app on airplane wifi, flying from Germany to SF. That was fun.
Clippy sounds like a dream for staying focused! I’m really curious if the local LLMs are fast enough to keep up with real-time work without any lag. Since it pastes directly into my active window, I wonder if there’s a way to double-check or edit the response before it actually goes in. How well does it handle voice commands in noisy environments, and is setting up those integrations like Linear and Google Calendar really as simple as it sounds? Also, since everything is local, I’d love to know how much it impacts the computer's battery and performance!
I need clippy to knock on the screen to give me the full member berries
Congrats on the launch! Love the local-first, context-aware Clippy for frictionless in-app assistance.
People can already get system-wide dictation and “polished output” from existing tools—what’s the clearest switching trigger you’ve seen for Clippy (e.g., in-place editing, context memory, local-first privacy, tool actions), and what’s your 30‑second proof that it’s meaningfully better in daily use?
This is actually quite useful, could definitely unlock a lot of productivity in my day to day tasks 🚀 Thanks for sharing @maximilian_prokopp 🔥
Eliminating the 'copy-paste dance' between the browser and the text field is a huge UX win. It’s one of those friction points we just got used to, but didn't realize how annoying it was until a solution like Clippy arrived! The Linear and Google Calendar integration via voice is a game changer for productivity. My question: Which local model are you currently using for the 'context-aware' part? And is there a way for users to swap in their own models (like through Ollama)? Does Clippy 'see' the context of the specific app I'm in (like a code editor vs. an email client), or is it based on the text I've already typed in the field?
Coding 90% of the app on airplane Wi-Fi is the ultimate builder flex, thats an incredible story, Max.The liquid glass aesthetic in Figma really paid off and It’s clear that coming from a film background gave you a unique eye for the visuals the app looks beautiful! Only one question, are you planning to keep it strictly for macOS, or is a mobile companion app in the cards?
Is the extension safe? :D Lately I was trying more extensions and LinkedIn started accusing me I am using some not good tools so they restricted my account for a while. Because this looks like a cool tool but I do not want to risk anything :D
Context-aware “paste into active field” plus voice-triggered MCP tool calls can go sideways fast at scale: wrong-window injection, secure fields, and unintended side effects (Calendar, Linear) from misheard intent.
Best practice is capability-based tool permissions with explicit usaer-confirm steps and audit logs, plus OS-level allowlists and secure-field detection; for local dictation, streaming whisper.cpp-style on-device ASR with VAD keeps latency predictable and private
How do you sandbox MCP execution and prevent writes unless the focused app and target field pass an allowlist, especially around password or secure input contexts?
Hi Product Hunters 👋,
I'm Max,
I am coming from a non-technical background, from film production, and a bit more than a year ago I started building things that I like, that are useful for me, and that are fun.
I built Clippy because I wanted to have a cool way to dictate with local models and later I realised the whole GPT context copy and direct pasting possibility. And then ideas started storming with linear and co. I have a newer version with control of apple calendar and reminders, but thats still a work in progress.
And I chose Clippy for a vintage cute touch.
I did all the visuals myself in Figma and experimented a lot with liquid glass and local models.
Ah and as a fun fact - I coded around 90% of the app on airplane wifi, flying from Germany to SF. That was fun.
Hope you like it too.
Best,
Max