TR

Omnara Launches Agentic IDE to Run Claude Code and Codex Anywhere

Omnara, a Y Combinator S25 startup, has unveiled a groundbreaking web and mobile IDE that enables developers to run Claude Code and Codex agents locally or in the cloud, with seamless session continuity across devices—including voice-enabled hands-free interaction.

calendar_today🇹🇷Türkçe versiyonu
Omnara Launches Agentic IDE to Run Claude Code and Codex Anywhere

Omnara, a Y Combinator Spring 2025 startup founded by Kartik, Ishaan, and Christian, has launched an innovative agentic integrated development environment (IDE) that transforms how developers interact with AI coding assistants like Anthropic’s Claude Code and OpenAI’s Codex. Unlike remote agent platforms that operate in isolated cloud VMs, Omnara runs agents directly on the user’s machine—whether a laptop, desktop, or remote server—while providing a unified web and mobile interface to monitor, guide, and continue coding sessions from anywhere. The platform’s architecture eliminates the need for SSH, port forwarding, or complex tunneling by using a lightweight, outbound WebSocket daemon that maintains a secure, authenticated connection to Omnara’s servers, ensuring enterprise-grade security and ease of deployment.

One of Omnara’s most significant innovations is its ability to persist agent state across devices and environments. When a user’s local machine goes offline, Omnara automatically resumes the session in a hosted remote sandbox, syncing code changes via Git commits at each interaction. This ensures that whether a developer is working from their home office, a coffee shop, or a mobile device, the agent picks up exactly where it left off. While environment parity in the sandbox isn’t perfect, the team notes that missing dependencies are typically resolvable by prompting the agent to install them—an approach that mirrors how developers naturally troubleshoot in real-time.

Perhaps the most unexpected breakthrough came with the addition of a voice agent. Initially conceived as a novelty feature for developers on the go, voice interaction has proven unexpectedly effective. Users report that speaking aloud to the agent—during walks, commutes, or while multitasking—leads to more detailed, iterative, and ultimately higher-quality code plans than typed prompts. The redundancy and natural elaboration inherent in speech appear to align better with how AI agents interpret intent, reducing ambiguity and improving task completion rates. "Talking through an idea feels more intuitive than typing a prompt," said one early adopter. "It’s like having a pair programmer who never gets tired."

Installation is straightforward: users run a single curl command to install the Omnara daemon within any Git repository. Once activated, the agent begins running locally, and sessions instantly appear in the web and mobile apps. The platform supports both Claude Code and Codex, leveraging users’ existing API subscriptions—meaning Omnara charges only for its infrastructure and session management, not for AI tokens. Pricing is tiered: free for 10 sessions per month, $20/month for unlimited sessions.

While competitors like Devin and Codex Web operate entirely in the cloud, Omnara’s hybrid approach—local execution with cloud fallback—resonates with developers concerned about data privacy, latency, and control over their codebase. The company’s decision to prioritize GUIs over CLI/TUI interfaces for mobile and web access reflects a broader industry shift toward human-centered AI tooling. Early feedback from Hacker News users has been overwhelmingly positive, with many praising the seamless cross-device experience and the surprising utility of voice interaction.

As AI agents evolve from assistants to true collaborators, platforms like Omnara are redefining the boundaries of software development. By decoupling the agent’s execution from the user’s physical location, Omnara doesn’t just extend productivity—it reimagines it. The company is now exploring integrations with CI/CD pipelines and expanding sandbox environments to support more complex dependency graphs, signaling that this is only the beginning of a new era in AI-augmented coding.

recommendRelated Articles