A desktop coding agent with free access to frontier models.
TypeScript
const home = `# OmniRush.ai
> The open-source Claude Cowork alternative. Chat on files, use skills, schedule tasks, automate a browser, and run on any model — plus an MCP gateway for your whole team.
## What it is
- Free, open-source desktop app for macOS, Windows, and Linux
- Any model or provider: Claude, GPT, Gemini, Mistral, local models, and 50+ providers
- Chat on files, browser automation, scheduled tasks, skills, and Anthropic-compatible plugins
- OmniRush.ai Connect: one MCP gateway URL for org-wide skills, servers, roles, and policies
- OmniRush.ai Web in alpha; central management and private deployment options for teams
- Coming soon: Dispatch (assign tasks from your phone) and live artifacts (auto-refreshing dashboards)
## Primary calls-to-action
- **Download for free** — [Desktop](https://github.com/omnirush-ai/omnirush-gui/releases/latest)
- **Open in your browser** — [OmniRush.ai Web](https://app.omnirushlabs.com)
- **Team plans** — [Pricing](https://omnirush.ai) (first 5 seats free, then \\$10 per seat/mo)
- **Sign in to the hosted workspace** — [Cloud](https://app.omnirushlabs.com)
- **SSO / audit / procurement** — [Enterprise](https://omnirush.ai)
- **Docs** — [README](https://github.com/omnirush-ai/omnirush-gui#readme)
- **Migrate from Claude Cowork** — [Migration guide](https://github.com/omnirush-ai/omnirush-gui/blob/main/packages/docs/start-here/migrate-from-claude-cowork.mdx)
## How it compares
- **vs Claude Cowork** — feature parity without model or vendor lock-in, plus an MCP gateway usable from any compatible client
- **vs Codex** — general knowledge work on files (not only coding), model and provider agnostic
- **vs ChatGPT Desktop** — agents act on local files and tools (MCP, plugins, skills) with guardrails; the setup is shareable and self-hostable
## FAQ
### What is OmniRush.ai?
A free, open-source desktop app (macOS, Windows, Linux) for doing work with AI agents on your own files. Built on OpenCode; an open-source alternative to Claude Cowork and Codex.
### Is OmniRush.ai free?
Yes — the desktop app is free and open source with bring-your-own keys. Team Starter includes your first 5 seats free, then \\$10 per seat/mo; Enterprise is custom.
### Which models does it support?
Any model OpenCode supports: OpenAI, Anthropic, Google, local models — 50+ providers.
### Does it send files to the cloud?
No. Desktop mode keeps files local; prompts go directly to your chosen LLM provider. Cloud workers are optional.
## For agents
- Agent skills index — \`/.well-known/agent-skills/index.json\`
- llms.txt — \`/llms.txt\`
- API catalog (RFC 9727) — \`/.well-known/api-catalog\`
- Sitemap — \`/sitemap.xml\`
Backed by Y Combinator.
`
const pricing = `# OmniRush.ai pricing — free, team, and enterprise
> OmniRush.ai has three tiers: free open-source desktop, Team Starter with the first 5 seats free then \\$10 per seat/mo, and custom Enterprise.
## Solo — Free
- Open-source desktop app
- macOS and Linux downloads
- Bring your own provider keys
- Free forever
- CTA: [Get Started for free](https://app.omnirushlabs.com?mode=sign-up)
## Team Starter — \\$10 / seat / month
- First 5 seats free
- API access
- Extension Marketplace
- Bring your own LLM keys, distributed to your team
- CTA: [Start team plan](https://app.omnirushlabs.com/dashboard/billing)
## Enterprise — Custom pricing
- Everything in Team Starter
- SSO / SAML and SCIM provisioning
- Bring your own inference — self-hosted or private models
- Desktop policies and version controls — admins decide which providers, models, extensions, and app versions employees can use; the desktop app enforces it automatically
- Managed deployment — self-hosted in your environment or hosted by OmniRush.ai
- Custom skill development and MCP consulting
- Enterprise rollout support and custom commercial terms
- Existing organizations already using SSO or desktop policies keep full access (grandfathered)
- CTA: [Talk to us](mailto:[email protected]?subject=Enterprise)
Prices exclude taxes.
`
const enterprise = `# A privacy-first alternative to Claude Cowork for your organization
> The open-source Claude Cowork alternative — self-hosted, permissioned, and compliance-ready. SSO, audit, custom deployment, and procurement support.
## What Enterprise includes
- SSO / SAML integration and SCIM provisioning
- Desktop policies and version controls — guardrails for providers, models, extensions, and app versions, enforced by the desktop app
- Managed deployment — self-hosted in your environment or hosted by OmniRush.ai
- Custom skill development for your team's workflows
- MCP consulting — connect internal data sources and tools as MCP servers
- Enterprise rollout support and custom commercial terms
- Named security contact and incident response
## Deployment models
- Self-hosted desktop app — data stays local, bring your own keys
- Cloud workers — managed by OmniRush.ai, sandbox infrastructure via Daytona (EU)
## Next step
- [Email us](mailto:[email protected]?subject=Enterprise)
- [Security Review](https://omnirush.ai) — data handling, subprocessors, and incident SLA
- See [Pricing](https://omnirush.ai) for tier comparison
`
const trust = `# Trust & Security
> How OmniRush.ai handles data, what subprocessors are involved, and how to reach the security team.
## Key facts
- **Deployment** — self-hosted desktop app on your machines
- **Data storage** — local-only, nothing leaves your machine in desktop mode
- **LLM keys** — bring your own, sent directly to your provider
- **Telemetry** — none in desktop mode; opt-in feedback only
- **Incident SLA** — 72hr notify, 3-day ack, 7-day triage
- **Subprocessors** — 5 named vendors (cloud & website only)
## Data handling
| Data type | Self-hosted | Cloud |
|---|---|---|
| Source code | Local only | Accessed at runtime via your LLM provider; not stored |
| LLM API keys | Local keychain / env vars | Held by your LLM provider, not by OmniRush.ai |
| Prompts & responses | Local only | Sent to your LLM provider; not logged by OmniRush.ai |
| Usage telemetry | None | Anonymous via PostHog; can be disabled |
| Authentication | Your SSO / SAML | Google or GitHub OAuth |
## Subprocessors
- PostHog — analytics (US/EU)
- Polar — billing (US)
- Google — OAuth (US)
- GitHub — OAuth (US)
- Daytona — cloud sandbox infrastructure (EU)
## Security contact
[email protected] (subject: security)
`
const glm52 = `# GLM 5.2 is now in OmniRush.ai — with 2x usage
> GLM 5.2 is available through OmniRush.ai Models, and we're doubling your usage so you can run real agent work on an open model at a fraction of the cost.
## What's new
- **GLM 5.2 in OmniRush.ai Models** — managed OSS model access with 2x usage, no keys required
- **Run your day from chat** — tasks organize into In progress / Done / Requires attention; move them by asking
- **Split screen** — two windows side by side, less tab-switching
- **Voice mode** — control the OmniRush.ai UI by voice
- **Advanced analytics on OmniRush.ai Cloud** — usage, activity, and team behavior in one view
## How it works
1. **Sign up** — [Get Started for free](https://app.omnirushlabs.com?mode=sign-up&intent=models)
2. **Subscribe** — OmniRush.ai Models at $10/user/mo includes GLM 5.2 with 2x usage
3. **Open the app** — switch to GLM 5.2 from the model picker
## What to try first
Open OmniRush.ai, switch to GLM 5.2, and ask the chat to organize your tasks.
## Links
- [Try GLM 5.2 in OmniRush.ai](https://app.omnirushlabs.com?mode=sign-up&intent=models)
- [Download the app](https://github.com/omnirush-ai/omnirush-gui/releases/latest)
- [Full changelog](https://github.com/omnirush-ai/omnirush-gui/blob/main/packages/docs/changelog.mdx)
`
const download = `# Get Started with OmniRush.ai
> Create a free OmniRush.ai Cloud account first, then use the guided desktop app access flow.
## Start here
- [Get Started for free](https://app.omnirushlabs.com?mode=sign-up)
- Create or select your workspace.
- Follow the Cloud app's desktop app access flow.
## Supported platforms
- macOS
- Windows
- Linux
## After signing up
Once the desktop app is running, use the [workspace-guide skill](https://omnirushlabs.com/.well-known/agent-skills/workspace-guide/SKILL.md) for first-run orientation.
`
const connect = `# OmniRush.ai Connect
> The MCP gateway for your whole org. Add a server or skill once, then share it with every teammate and agent through one URL.
- Org-level authentication, roles, allowlists, and audit apply to every call
- Works in OmniRush.ai and any MCP-compatible client
- First 5 seats are free
- [Get started free](https://app.omnirushlabs.com?mode=sign-up)
- [Read the docs](https://github.com/omnirush-ai/omnirush-gui#readme)
`
const cloud = `# OmniRush.ai Cloud
> The dashboard for running OmniRush.ai across your organization.
- Provision model providers centrally
- Deploy skills and MCP servers to every seat
- Manage members, policies, usage, and audit
- OmniRush.ai Web and the Connect MCP gateway are built in
- [Get started free](https://app.omnirushlabs.com?mode=sign-up)
- [Explore Connect](https://omnirush.ai)
`
export const agentMarkdown: Record<string, string> = {
"/": home,
"/connect": connect,
"/cloud": cloud,
"/pricing": pricing,
"/enterprise": enterprise,
"/download": download,
"/trust": trust,
"/glm-5.2": glm52,
}
export const agentMarkdownRoutes = Object.keys(agentMarkdown)