Cord
Cord is a cloud-based remote coding platform that lets developers run Claude Code, Codex, and OpenCode AI agents in isolated, ephemeral sandboxes. Users can launch agent sessions from Slack, Linear, or GitHub tickets, with each sandbox providing a full dev environment, live preview URL, and persistent storage for dotfiles and test artifacts. The platform operates on a Bring Your Own Key model, using users' existing AI subscriptions or API keys to avoid double-paying and vendor lock-in.
- Developer Tools
- AI Agents
- Artificial Intelligence
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Developers face friction when running AI coding agents locally, including hardware constraints, environment inconsistencies, and the inability to run multiple agent sessions in parallel. Coordinating agent work across teams and reviewing outputs without dedicated infrastructure slows down shipping and creates context-switching overhead between coding, communication, and review tools.
Solution
Cord provides a cloud-based remote coding platform where developers can run Claude Code, Codex, and OpenCode agents in isolated, ephemeral sandboxes. Users log in with GitHub, connect repositories, and launch agent sessions from Slack, Linear, or GitHub tickets; each sandbox spins up in seconds, runs a full dev environment, and suspends automatically when idle. A unique preview URL per sandbox lets developers or the agent itself verify changes live, and persistent storage preserves dotfiles, configuration folders, and test artifacts across sessions. The platform works entirely on a Bring Your Own Key model—users' existing Claude or Codex subscriptions or API keys are encrypted and used directly—eliminating double-paying and vendor lock-in.
Target Audience
Primary users are software developers and engineering teams who use AI coding assistants and need scalable, remote infrastructure to run agent-based workflows across multiple repositories and devices.
Features
- Cloud sandbox runtime supporting Claude Code, Codex, and OpenCode agents with Docker builds fully supported
- Bring Your Own Key model where Anthropic or OpenAI API keys are encrypted at rest and used solely for agent calls
- GitHub OAuth-based authentication with repository metadata and branch-level session isolation
- Live preview URLs generated per sandbox for verifying changes, sharing with teammates, or agent self-verification
- Persistent user storage that carries dotfiles, .claude directories, and test artifacts across sandbox sessions
- Parallel session execution with automatic idle suspension and instant resume to minimize compute costs
- Voice message initiation and mobile access for starting agents and reviewing diffs on the go