Free & Open Source  ·  macOS + Windows

AI-powered text
enhancement, anywhere.

Select any text, press a shortcut — OmniKey enhances prompts, fixes grammar, and runs custom AI tasks in any app, on any screen.

One-line install 
curl -fsSL https://omnikeyai.ca/install.sh | bash

macOS 13+  ·  Windows 10/11 x64  ·  Free to use

Works with
OpenAI · Anthropic · Google Gemini · NVIDIA Nemotron

See it in action

Any application on your computer
Selected text
fix my code it's not working
⌘E Enhance Prompt
AI Enhanced & pasted back
Please review the provided code and identify the issue causing it to malfunction. Explain what's going wrong, then provide a corrected version with brief comments explaining the key fix.
OmniAgent

One prompt away from any specialist

OmniAgent is a generalist with terminal access — pair it with the right MCP servers and it transforms into a coder, researcher, ops engineer, or data analyst on demand.

Agent Chat Page

Open the chat and ask anything

Head to the Agent page in the desktop app, write a base prompt that defines the role, and start chatting. OmniAgent keeps the full conversation context — ask follow-ups, refine, branch off into new tasks without losing the thread.

Invoke from any screen

Run a quick task anywhere

Don’t want to leave the app you’re in? Trigger OmniAgent inline with a keyword.

  1. 1
    Write the prompt
    @omniAgent find all TODOs in this project and summarise them
  2. 2
    Select the text
    Highlight what you just typed in any app.
  3. 3
    Press ⌘T / Ctrl+T
    The agent picks up your selection and runs it.
Simple by design

How OmniKey works

Three steps, zero friction. Works in every app on your computer.

01

Select any text

Highlight text in any application — emails, documents, code editors, chat apps.

02

Press a shortcut

Use ⌘E, ⌘G, or ⌘T to trigger the AI enhancement.

03

AI enhances in-place

The result is automatically pasted back, replacing your selected text. No copy-paste needed.

Keyboard shortcuts

Three powerful commands

Each shortcut maps to a specific AI capability, available in any app on your screen.

⌘E

Enhance Prompt

Rewrite selected text with better clarity, structure, and tone — ideal before sending to any AI tool.

⌘E macOS  ·  Ctrl+E Windows
⌘G

Grammar & Clarity

Fix spelling, grammar, and readability without changing the core meaning of your text.

⌘G macOS  ·  Ctrl+G Windows
⌘T

Custom Task + Agent

Apply your own saved instructions or run the AI agent with real-time web search and terminal access.

⌘T macOS  ·  Ctrl+T Windows
Advanced usage

Unlimited flexibility with @ commands

Prefix your text with @ to give OmniKey specific instructions on the fly.

⌘E

Enhance a Prompt

Select any text in any app and press ⌘E (macOS) or Ctrl+E (Windows). OmniKey rewrites it with better clarity, structure, and tone — ideal before sending to an AI tool.

⌘G

Fix Grammar & Clarity

Select text and press ⌘G / Ctrl+G. OmniKey corrects spelling, grammar, and readability without changing your meaning.

⌘T

Run a Custom Task

Set up your own task instructions in the Task Instructions tab in the menu bar app. Then select any text and press ⌘T / Ctrl+T to apply them.

@

Ask OmniKey Anything

Start your text with @omnikeyai followed by any instruction. Select it and press any OmniKey shortcut — OmniKey takes action based on what you write and pastes the result in place. 

@omnikeyai write this text into valid markdown format

Powered by your configured provider — OpenAI, Anthropic, Google Gemini, or NVIDIA Nemotron.

🤖

Run the AI Agent

Start your instructions with @omniAgent, select them, and press ⌘T / Ctrl+T. The agent gathers context from your terminal and the web, combines your saved task instructions if configured, then completes the task and pastes the result. 

@omniAgent find all TODO comments in my project and summarise them

Supported web search providers: DuckDuckGo, Serper, Brave Search, Tavily, SearXNG.

Everything included

Packed with features

Multiple LLM Providers

Choose between OpenAI, Anthropic, Google Gemini, or NVIDIA Nemotron (open weights, served via NVIDIA NIM) as your AI backend. Configure your preferred provider during onboarding or at any time via the CLI.

MCP Servers

Plug in any MCP (Model Context Protocol) server — via stdio, HTTP, or SSE transport — from the macOS app, Windows app, or CLI. The agent automatically discovers and calls those tools during task execution.

Web Search

Give the agent real-time web context. Supported providers: DuckDuckGo, Serper, Brave Search, Tavily, SearXNG.

Prompt Enhancement

Press ⌘E / Ctrl+E to improve the clarity, structure, and tone of selected text — ideal before sending to any AI tool.

Grammar & Clarity Fix

Press ⌘G / Ctrl+G to correct spelling, grammar, and readability without changing the core meaning of your text.

Custom Tasks

Press ⌘T / Ctrl+T to apply your own saved task instructions to the selected text. Configure them in the Task Instructions window from the menu bar.

Any device, anywhere

Telegram Integration

Interact with @omniAgent and receive notifications directly in Telegram from your phone or any device. Run the daemon and the Telegram client on your machine, leave it running, and drive entire agent sessions from Telegram. Use /cmd to start or resume a task via a guided wizard, /task to check progress or fetch the last result, and /stop to abort a running session. The bot runs as a persistent background daemon (launchd on macOS, NSSM on Windows) managed entirely through the CLI.

Get started with one command 

omnikey telegram start

On first run the CLI prompts for your bot token (from @BotFather) and chat ID, validates the token against the Telegram API, and saves both to ~/.omnikey/config.json. For the full setup guide see the Telegram README.

macOS & Windows

Authenticated Browser Session Reading

OmniKey reads content directly from your live browser tab — including pages that require login. The @omniAgent can work with private dashboards, internal docs, and authenticated tools without any copy-pasting.

Works with: Chrome, Brave, Edge on Windows  ·  Chrome, Brave, Edge, Arc, Vivaldi, Opera, Chromium, Safari on macOS.

Recommended: one command does everything 

omnikey grant-browser-access

The CLI detects your installed browsers, creates a dedicated Omnikey debug profile stored at ~/.omnikey/browser-debug-profiles/, finds a free debug port, registers a permanent login startup entry, and verifies the connection — all in one guided flow.

Because a fresh profile is used, you sign in to your accounts once inside it. Your main browser profile is never touched, and the debug profile persists between sessions.

After setup, reopen the browser with its Omnikey debug profile any time using: 

omnikey browser open

Manual setup

Chrome, Brave, Edge, Arc, Vivaldi, Opera, Chromium

  1. In the menu bar go to View → Developer → Allow JavaScript from Apple Events.
  2. Make sure it is checked, then restart the browser.

Safari

  1. Go to Safari → Settings → Advanced and enable "Show features for web developers".
  2. Then go to Develop → Allow JavaScript from Apple Events and make sure it is checked.

No separate profile or port needed — Omnikey reads your existing open tabs directly. This is a permanent one-time setting.

Automation

Scheduled Jobs

Run prompts automatically on a recurring schedule or at a one-time date and time.

App

Use Scheduled Jobs in the Desktop App

Open Scheduled Jobs from the OmniKey menu bar (macOS) or system tray (Windows). You can:

  • Create recurring jobs using weekday/day selection and time
  • Create one-time jobs for a specific date and time
  • Run jobs immediately with Run Now
  • Enable/disable jobs with Activate and Deactivate
  • Inspect the last execution with Last Run
CLI

Manage Scheduled Jobs from the CLI

Use the schedule command group to add, list, remove, and trigger jobs: 

omnikey schedule add
omnikey schedule list
omnikey schedule remove
omnikey schedule run-now <job-id>

The add flow is interactive and supports multiline prompt input, recurring cron schedules, and one-time jobs.

Extensibility

MCP Servers

Extend the OmniKey agent with any external tool by connecting MCP servers over stdio, HTTP, or SSE.

App

Manage MCP Servers in the Desktop App

Open MCP Servers from the OmniKey menu bar (macOS) or system tray (Windows). From there you can:

  • Create a new server — enter a name, description, and choose a transport: stdio, http, or sse
  • For stdio: supply the command, arguments, and optional KEY=VALUE environment variables
  • For http / sse: supply the endpoint URL and optional KEY=VALUE headers
  • Toggle a server on or off with Enable / Disable
  • Refresh a server connection with Refresh
  • Edit server details or delete a server at any time
CLI

Manage MCP Servers from the CLI

Use the mcp command group to register, inspect, and control servers: 

omnikey mcp add
omnikey mcp list
omnikey mcp remove
omnikey mcp toggle <id>
omnikey mcp update <id>

add and update are interactive prompts that ask for name, description, transport (stdio, http, sse), enabled flag, and the transport-specific fields.

Remote access

Telegram Integration

Run the daemon and the Telegram client on your machine, leave it running, and handle entire OmniAgent sessions from Telegram on any device.

Chat with OmniAgent from Telegram

Once the daemon is running, send commands to your bot from any device. All commands are accepted only from your configured chat ID.

  • /cmd — guided wizard to start or resume an agent session
  • /cmd --verbose — stream shell output, web calls, and MCP events as they happen
  • /task — check progress of a running session, or fetch the last result
  • /stop — abort the running agent session
CLI

Manage the Telegram Daemon

The bot runs as a persistent background service — launchd on macOS, NSSM on Windows. On first run it prompts for your bot token and chat ID, validates them, and saves them to ~/.omnikey/config.json. 

omnikey telegram start    # install and start
omnikey telegram status   # check if running
omnikey telegram logs     # tail stdout + stderr
omnikey telegram restart  # restart after a config change
omnikey telegram stop     # stop the daemon

For setup instructions — creating a bot with @BotFather and finding your chat ID — see the Telegram README.

Quick setup

Up and running in minutes

Four simple steps to get OmniKey working on your machine.

1

Install the CLI

Install the OmniKey CLI:

brew install GurinderRawala/omnikey/omnikey-cli
2

Onboard & configure your AI provider

Run the interactive setup to pick your LLM provider and enter your API key:

omnikey onboard

Supports OpenAI, Anthropic, Google Gemini, and NVIDIA Nemotron. You can also optionally configure a web search provider. Supported web search providers: DuckDuckGo, Serper, Brave Search, Tavily, SearXNG.

3

Start the daemon

Register OmniKey as a persistent background service that survives system restarts: 

omnikey daemon

On macOS this registers a launchd agent. On Windows it registers a Task Scheduler task.

4

Download & launch the app

Download the app above for your platform, install it, and you're ready. Select any text in any app and press a shortcut. You can manage task instructions and view the manual from the menu bar / system tray icon.

OmniKey AI

Download OmniKey AI

Free to use. Open source. Works on macOS and Windows.

Download for macOS Download for Windows

macOS 13+  |  Windows 10/11 x64