SuperHQ

About SuperHQ

SuperHQ is a developer-focused tool for running AI coding agents inside isolated microVM sandboxes. It orchestrates Claude Code, Codex, and Pi agents while keeping API keys out of the guest environment via a local auth gateway. Launched recently, it targets safer agent-driven code workflows.

Review

SuperHQ delivers per-agent microVM isolation, mounting projects into a Debian environment and directing agent writes to a tmpfs overlay so the host filesystem remains untouched. The app provides a unified diff workflow to review and accept or discard changes, and supports using a ChatGPT subscription directly for Codex and Pi without exposing API keys.

Key Features

• One microVM per agent with a full Debian environment for true isolation.
• Project mounts with writes routed to a tmpfs overlay, protecting the host filesystem.
• Local auth gateway that swaps API keys on the wire so keys never enter the VM.
• Unified diff view to inspect and approve or reject agent-made changes.
• Support for multiple coding agents (Claude Code, Codex, Pi) and optional use of a ChatGPT subscription for some agents.

Pricing and Value

SuperHQ is currently listed as free on launch. Its main value proposition is improved safety and auditability for workflows that use AI coding agents: isolated environments reduce the risk of accidental host modifications, while the diff-based review model gives developers control over agent outputs. For teams or individuals who run untrusted agents or need stronger separation between tools and local code, the tradeoff is likely worth it, especially at no cost for early adopters.

Pros

• Per-agent microVMs provide a higher level of isolation than terminal wrappers or simple worktree approaches.
• API keys are kept out of guest VMs via a local auth gateway, improving secret safety.
• Unified diff workflow simplifies review and recovery from unwanted changes.
• Supports multiple popular coding agents and can leverage an existing ChatGPT subscription for some integrations.

Cons

• Platform support is limited at launch (currently MacOS only; Linux and Windows ports are planned).
• Early-stage release may have rough edges and a smaller community for support and extensions.
• Some users have reported minor compatibility quirks with specific agents (for example, interface or cursor behavior with Pi).

Overall, SuperHQ is best suited for developers and teams who want a safer, reviewable way to run AI coding agents locally and are comfortable trying an early-release tool on MacOS. It's a strong fit for security-minded workflows and experimental automation where preserving the host environment and controlling agent edits are priorities.

Open 'SuperHQ' Website