Content stays on your Mac
Dictation, transcripts, meeting audio and local indexes are processed and stored on your machine.
Included at launchWhy it works
A clear PR description is the difference between a one-day review and a one-week review. Penwright makes that description cheap to write. Describe what changed, why it changed, how to test it, and let Penwright structure it for the reviewer.
You say
this pr fixes the bug where users couldnt sign up if they had an apostrophe in their last name the regex was too greedy
Penwright writes
## What Fixes a signup bug where users with an apostrophe in their last name were rejected. ## Why The surname validation regex was too greedy and treated `'` as invalid. ## How to test 1. Sign up with last name `O'Brien` 2. Confirm account is created.
In the wild
Three moments in GitHub where Penwright pays for itself.
The quick one
Click into a GitHub text field, hit the hotkey, speak. The clean text is in before you would have typed it.
The long one
Three minutes of speaking become structured text with paragraphs and punctuation. No typing cramp, no reformatting.
The recurring one
Standard replies live as snippets and expand from a spoken trigger. The dictionary keeps names, companies and acronyms spelled right.
Day one: setting Penwright up for GitHub.
- Install Penwright
Grab the .dmg from the download page, drop it in Applications, open.
- Pick your hotkey
Default is ⌘⇧Space. If Spotlight is in the way, set a conflict-free shortcut right away.
- Seed your dictionary
Drop in the names, companies and codenames you use every day. Penwright will get them right from day one.
- Dictate your first GitHub message
Open GitHub, click into a text field, hold the hotkey, speak. Release. The cleaned version lands in the field.
Frequently asked
- Does Penwright work in GitHub?
- Yes. Penwright types wherever your cursor lands in a text field, including GitHub. There's no separate GitHub integration; Penwright uses macOS' standard text insertion.
- Are my GitHub messages sent to the cloud?
- No. Transcription and cleanup run locally on your Mac, Metal-accelerated. If you add your own Claude or OpenAI key in Settings, then yes, but only when you explicitly activate it, and a visible CLOUD badge stays on while it is.
- Does Penwright understand GitHub-specific jargon?
- Decent out of the box, perfect with your own dictionary. Add company names, people names, codenames once in Settings. Penwright will type them correctly into GitHub from that moment on.
- Do I need a fast internet connection?
- No. Once the models are installed, Penwright works offline. Ideal for travel, airports, long calls.
In GitHub, typing is the real tax.
Three to four times faster with voice. The math gets uncomfortable fast.
Typing40WPM
Average typing speed for a confident keyboard user.
Voice with Penwright150WPM
Sustainable speech rate. No cloud round-trip.
Privacy
Privacy by architecture, not by policy.
Penwright does not need your spoken content to provide the product. Audio, transcripts and local indexes stay on your Mac, and you can verify the network traffic yourself.
Data minimisation by default
Penwright's servers handle accounts, licences and billing, not the audio or transcripts you create on your Mac.
Included at launchNo audio or text upload endpoint
The Penwright API has no route for your dictation audio or transcripts. A server cannot leak content it never receives.
Included at launchVerify every claim
Use any network monitor you trust. The security page documents the traffic you should, and should not, see.
Included at launchRelated apps
Cursor
Dictate long prompts and code intent directly into Cursor. Stop typing, start shipping.
VS Code
Comments, commit messages, JSDoc, dictate the boring parts of writing code.
ChatGPT
Talk to your AI like you talk to a colleague. Long, structured prompts without the cramp.
Gmail
Reply to a full inbox in a fraction of the time. Speak the response, send the clean version.