The meta agent harness
for leverage addicts
We make it fun to run 100+ agents, coding or otherwise.
Any harness, any AI, all in an interface optimized for people who love getting things done.
Supports Claude and Codex subscriptions + any harness.
Fair warning: you'll never be able to work the old way again.
A hundred agents.
One calm inbox.
No more monitoring terminals: agents only alert you when a decision actually needs you.
Fifteen seconds. One decision. Back to quiet.
Be the boss, not the babysitter.
Every update,
in plain English.
Every update, rewritten in plain English.
Skim it in seconds. Decide. Move on.
I have completed the pricing page implementation. Summary of changes:
1. Created PricingTable.tsx (312 lines) with three tiers, an annual/monthly toggle and a feature comparison matrix. Tier data flows through usePricingTiers() reading pricing.config.ts, so copy edits never touch components. The toggle persists its state in localStorage under pricing.billing and falls back to monthly when the key is absent or malformed.
2. Added PricingFaq.tsx and the /pricing route in router.tsx; the header nav now links to it. Focus order and ARIA labels verified with the keyboard-nav suite. The FAQ accordion uses a roving tabindex and announces expanded state via aria-expanded.
3. Tests: 128 passing (12 new: tier toggle, currency formatting, keyboard nav, empty state). Lint and typecheck are clean. Bundle delta is +11.4 KB gzipped, under the budget. Snapshot tests were updated where the DOM structure changed; no logic changes were required for the existing checkout flow.
4. Note: the Stripe key in .env.example is still the test key. I can switch it to the live key from your vault as part of the deploy, but I did not want to do that without asking. There is also a TODO in PricingTable.tsx about extracting the comparison matrix once the enterprise tier copy is finalized.
5. Deployment considerations: the route is statically prerendered, so a CDN purge of /pricing will be needed after deploy; the existing deploy script already handles purges for new routes, but I want to flag it in case the pipeline configuration differs in production.
Would you like me to (a) switch the key and deploy to production now, (b) open a PR for review first, or (c) hold? Let me know how you would like to proceed and I can continue from here.
Describe it.
Watch it ship itself.
Not a demo. Not a snippet.
Our software factory ships real, working apps and high-quality PRs — regardless of your skill level.
Including the page you’re reading.
- Plan
- Build
- Test
- Review
- Ship
Prompt in. Merged PR out.
A built-in dev pipeline plans, builds, reviews and tests every change, with approval gates where you want them. Review the diff, accept or reject, and commit without opening a terminal.
Set it once.
It runs itself.
Describe an automation — it builds, tests, and schedules itself.
Quick replies. Cron jobs. Autoresponders. AI filters.
Omniscio even runs Omniscio.
Your whole fleet,
in your pocket.
Kick off new agents, approve their work, and run the whole fleet — the full power of the desktop app, in your pocket.
Be 100× more productive — even from bed.
Any model.
All at once.
Pick the best AI for the job — or ask them all at once.
Any model, any harness — Claude Code, Codex, OpenCode, Devin, Pi, and Hermes all supported.
Manage your agents easily regardless of platform.
Fast as thought.
Fire a prompt from anywhere.
Hotkeys throughout.
You’ll breeze through your work.
Your keys.
Your rules.
Local-first. Export everything, anytime.
Hard spend caps — nothing runs away.
Nothing ships without your yes — and undo is one tap.
All work is becoming agent work.
We’re rebuilding the apps you pay for, one after another. One subscription instead of twenty.
See the full walkthrough
Got questions?
Looking for a how-to? Browse the full docs.
What is Omniscio?
The meta agent harness: one app that runs 100+ AI agents at once, coding or otherwise, on any model and any harness. Sessions that need you sort to the top, so you answer and move on.
Which models and harnesses does it work with?
Any model, any harness — Claude Code, Codex, OpenCode, Devin, Pi, and Hermes are all supported.
Does it run on Windows, Mac and Linux?
Yes. There is a Windows installer, a macOS build for Apple silicon and Intel, and a Linux AppImage.
Do I need to know how to code?
No. Describe what you want in plain English — the agents handle the rest.
Do I need my own AI account?
Yes. Your agents run on the AI account you already use: your Claude subscription or Anthropic API key, or the account behind Codex or any other supported tool. Omniscio does not include AI usage.
What does it cost?
There are two plans. The Free plan covers 8 projects, and Pro is $20/mo. Your AI usage is billed by your own provider, not by us. See pricing.
Can I keep my AI spend under control?
Yes. Every session shows its cost as it runs, and a per-session cap stops a session when it crosses your limit.
Can I check on my agents from my phone?
Yes. Turn on Remote Access to see and answer your sessions from your phone’s browser over a paired connection.
Is my data private?
Your code and sessions stay on your computer, and your agents talk straight to the AI provider you choose, on your own account. Your keys are encrypted on your device. Built-in helpers like Plain Speak send the text they work on to an AI provider, sometimes through our servers. We never use your prompts, code or conversations to train models, and we never sell your data. Details: Privacy Policy.
Still curious? Email support@omniscio.com.