Know which AI wrote
every line.

Per-line attribution for human, AI, and pasted code. Any editor. Any agent. Zero config.

$npm install -g @omnitype-code/cli
tools
Claude Code
Cursor
Windsurf
GitHub Copilot
Codeium
Aider
Codex
Cline
Gemini CLI
Continue
Antigravity
Amp
Claude Code
Cursor
Windsurf
GitHub Copilot
Codeium
Aider
Codex
Cline
Gemini CLI
Continue
Antigravity
Amp
Claude Code
Cursor
Windsurf
GitHub Copilot
Codeium
Aider
Codex
Cline
Gemini CLI
Continue
Antigravity
Amp
languages
TypeScript
Python
Go
Rust
Java
C++
Ruby
Swift
Kotlin
PHP
C#
Dart
Hindi
Arabic
Korean
SQL
Scala
Elixir
Haskell
Zig
TypeScript
Python
Go
Rust
Java
C++
Ruby
Swift
Kotlin
PHP
C#
Dart
Hindi
Arabic
Korean
SQL
Scala
Elixir
Haskell
Zig
TypeScript
Python
Go
Rust
Java
C++
Ruby
Swift
Kotlin
PHP
C#
Dart
Hindi
Arabic
Korean
SQL
Scala
Elixir
Haskell
Zig
Features
Per-line attribution
See exactly who wrote what.
Every line tagged as human, AI, or paste — live in the editor.
1
function processPayment(order) {
2
const validated = schema.validate(order);
3
if (!validated.ok) throw new Error();
4
return stripe.charge(order);
5
}
Human
AI
Paste
Model detection
Which model, which line.
Claude, GPT-4o, Gemini — automatically.
claude-sonnet-4-6Claude Code
48%
gpt-4oCursor
31%
gemini-2.0-flashGemini CLI
21%
Git-native
Attribution per commit, forever.
Stored as git notes. Survives rebase and squash.
a4f2c1add payment flow71%29%
b8e3d0fix validation bug94%6%
c1a9f2refactor auth module33%67%
Team dashboard
Real visibility without self-reporting.
AI use per developer, per model, per week. Your whole org in one view.
rishav
38% AI
alex
72% AI
priya
19% AI
sam
56% AI
Typing test
20+ languages. Live races. Assessments.
Code-accurate typing engine with Hindi, Arabic, Rust, Go and more.
CLI
Works outside VS Code too.
JetBrains, Neovim, any terminal editor.
$omnitype blame src/app.ts
$omnitype status
$omnitype hooks install
Open protocol
Any tool can integrate.
5 lines. One file. Claude Code ships it natively.
"file": "/project/src/index.ts"
CLI · Works in any editor

Not a VS Code user?
We got you.

MIT LicensemacOS · Linux · Windows
omnitype — zsh
@omnitype-code/cli
Install
$npm install -g @omnitype-code/cli
Any editor
Cursor, JetBrains, Neovim, Zed
Git-native
Hooks + git notes on every commit
Open protocol
Signal any model from any tool
Commands
omnitype daemon
Track any IDE in real time
omnitype blame src/index.ts
Per-line AI attribution
omnitype hooks install
Add git hooks to current repo
omnitype signal --model gpt-4o --tool aider
Signal active model manually
omnitype setup-vscode-hook
Enable all VS Code forks at once
omnitype status
Show active model and confidence
Provenance Engine

AI% doesn't tell
you if code is good.
That's the point.

A 95% AI commit can be great engineering or a liability. Provenance gives you the data to ask the right question, not a verdict.

Reviewers read with context
See which lines are AI-suggested before opening the diff. Same decision, better frame.
Managers spot patterns early
A shift from 20% to 90% AI overnight is a signal. Provenance surfaces it automatically.
Teams understand their tools
Know which repos, files, and developers rely on AI most. Answer questions you could not ask before.
COMMIT · 95% AI
Great engineering
AI 95%
Human 5%
src/queue.ts
97%AI scaffold
src/worker.ts
88%AI scaffold
fix/race.ts
0%← the actual insight

A dev reads a 400-line AI suggestion, spots the race condition on line 312, writes 8 lines to fix it, ships. The commit is 95% AI. The 5% is what matters.

Same number. Different context. Provenance surfaces the data. Your team supplies the judgment.

For Engineering Teams

Know how your
codebase gets built.

Every line classified at write-time: typed manually, AI-generated, or pasted. No surveys, no self-reporting. Objective data, like logs and metrics.

Overview
Profiles
Alerts
Composition shift detected — AI-generated lines up 40% this week
A
alex.chenAI-Assisted
Senior Eng
108 wpm
J
jordan.kAI-Heavy
Backend
61 wpm
S
sarah.mManual-First
Fullstack
94 wpm
T
tom.wrightAI-Assisted
Frontend
76 wpm
Manual
AI
Paste
  • Observability, not surveillance
    Provenance is infrastructure data, like CPU or latency. It tells you what is happening, not whether to trust your developers.
  • Code review with full context
    Reviewers know which lines came from AI before they open the diff. They still decide, but with the right information.
  • Answer the questions that matter
    "Which repos use AI most?" "Did that refactor shift our composition?" "Where is AI helping?" Now you can answer these.
  • Bulk hiring assessments
    Send one link to 2,000 candidates. AI usage, paste rate, and manual code ratio ranked automatically. No spreadsheets.
0
Setup required. Install and it just runs.
AI tools supported, tool-agnostic by design
100%
Passive collection. Devs just code normally.
1
Link to run a 2,000-seat assessment
Hiring Assessments

Test 2,000 developers
with one link.

Create an assessment room, share the link, and get ranked results automatically. Every paste event and manually typed line is tracked. Candidates see nothing. You see everything.

01
Create a room
Set language, time limit, and candidate count. Takes under a minute.
02
Share one link
Candidates join from any device. Paste events are flagged to you silently — candidates never see a flag, never know they were caught.
03
Get ranked results
Paste rate, manual code ratio, WPM, and accuracy — scored and ranked automatically. Download CSV instantly.
ASSESSMENT ROOM
Q4 Hiring Batch #3
JOINING
PARTICIPANTS — 14/100
PASTE ALERTS
visible to you only
Alerts appear when assessment starts
JOINED
14
PASTE EVENTS
SUBMITTED
ASSESSMENT PACKS — 10 ROOMS PER PACK
Need more seats or a recurring plan?
2,000
max seats/room
5d
assessment window
CSV
instant download
OmniType Attribution Protocolv1.1
Open Standard

Any AI tool can adopt it in 5 lines. Claude Code ships it natively.

~/.omnitype/active-model.json
Free for individuals, always

Start in minutes.

VS Code extension or CLI. Both free.

VS Code Extension
Cursor, Windsurf, any VS Code fork
Per-keystroke attribution
Paste detection
Live sidebar
Install from Marketplace
CLI
JetBrains, Neovim, Zed, any editor
Git hooks
Blame per-line
Daemon mode
$npm install -g @omnitype-code/cli
01
Install
Extension or CLI
02
Sign in
30 seconds
03
Code
Runs silently
04
See it
Live attribution
FAQ

Questions worth
answering honestly.

No fluff. How it works, what it sees, and what it doesn't.

It tells you — and your team — how your codebase actually gets built. Every line of code is tagged as manually typed, AI-generated, or pasted. That attribution travels with your commits so you always have an honest record of code authorship.

The VS Code extension watches how text enters your editor. Characters you type one by one are marked manual. A suggestion you accept from Copilot, Cursor, or Codeium is marked AI. Anything inserted with Ctrl+V is marked paste. No guessing — it tracks the actual input event.

Never. All the analysis happens locally on your machine. Only the resulting ratios are synced — something like '68% manual, 24% AI, 8% pasted' per commit. Your source code stays on your machine.

When you install the extension, it sets up git hooks in your workspace. Every time you commit and push, those hooks attach provenance metadata to the commit. Your org dashboard then shows a per-file, per-branch, per-contributor breakdown — built entirely from commit history.

Only your composition ratios and coding activity — not your actual code. Your manager sees things like how active you've been and what share of your commits were manual vs AI-assisted. They cannot see file contents, diffs, or anything you've written.

A live view of how your team's code actually gets written — per developer, per repo, per branch. You can see who is actively coding, how much AI tooling is being used across the team, and spot any unusual patterns. You only see members assigned to your team.

No noticeable impact. It reacts passively to editor events — there are no background scans, no file watchers, no polling. It only activates when you're actively editing.

The extension tracks files as you open and edit them. Files you haven't touched since installing it, or files generated by tools like npm or build pipelines, won't appear — they were never typed. Coverage grows naturally as you work.

Because they are. package-lock.json is written by npm — you never typed it. tsconfig.json often gets auto-filled by IntelliSense. The attribution is accurate. You can add generated files to .omnitype/ignore to exclude them from the dashboard entirely.

It's a deliberate practice tool — you practice typing real code in the languages you actually use, so your muscle memory matches your daily work. Think of it as a gym for your fingers. It's completely optional and separate from the provenance tracking.

The VS Code extension, provenance tracking, and individual dashboard are free. Team dashboards and org features require a plan. Assessment packs for hiring are available as one-time purchases.

Install the OmniType extension from the VS Code Marketplace, sign in with a free account, and open any project. The extension starts tracking immediately. Push a commit and your first provenance data appears in the dashboard within seconds.

Contact

Talk to us directly.

Questions, demos, integrations. You get a real answer.

Enterprise pricing
API access
Custom integration
General questions