Table of Contents
relayd
Most observability tools only let you watch your AI agents. relayd lets you command them—directly from your phone. A lightweight background daemon bridges your Local Codex or agent runtime to a secure mobile PWA. Receive real-time push notifications when tasks complete, review context-aware code diffs, and authorize next steps with a single tap.
Crucially, your code never leaves your machine. Unlike generic monitoring dashboards, relayd uses secure tunneling to keep your intellectual property local while giving you remote access. It is purpose-built for the agentic coding workflow: it’s not just an alert system, it’s a mission control for your local LLMs. Built by an indie dev, for indie devs. Your desk is now optional.
Features
- Mobile Mission Control: Push notifications for agent status and “human-in-the-loop” approval requests.
- Context-Aware Diffs: Review code changes and error logs directly on your mobile device before approving.
- Secure Tunneling: Zero-data-retention architecture ensures your codebase stays on your local hardware.
- One-Tap Actions: Instantly “Continue,” “Retry,” or “Abort” agent sessions remotely.
- Lightweight Daemon: Minimal resource footprint background process designed for local Codex/LLM runners.
How It Works
A lightweight daemon installed on your development machine creates a secure, encrypted tunnel to relayd’s mobile web app (PWA). This allows the mobile interface to receive status updates and send command signals (like approval or termination) back to your local agent without exposing your machine to the open web or uploading code to a third-party cloud.
Use Cases
- Commuter Coding: Monitor and approve long-running agent tasks while commuting or away from the keyboard.
- Remote Crisis Management: Pause or redirect runaway agents immediately upon receiving an error notification.
- Multi-Session Orchestration: Manage multiple local agent sessions simultaneously from a single mobile dashboard.
Pros & Cons
- Pros: Completely untethers developers from their desks; high privacy (local-first); optimized specifically for agent interaction patterns (diff review vs. raw logs).
- Cons: Niche focus (requires local agent setup); currently supports specific runtimes (e.g., Codex CLI); requires initial command-line configuration.
Pricing
- Free Tier: Includes 1 relay and 1 workspace.
- Pro ($5/mo): Unlimited relays and extended history.
How Does It Compare?
- vs. Lovable / Bolt / Devin: These are cloud-native platforms where code lives on their servers. relayd is for developers who run agents locally (e.g., Local Codex, OpenWebUI pipelines) and want the convenience of cloud tools without sacrificing data privacy or local compute control.
- vs. OpenWebUI / LibreChat: While OpenWebUI offers a great desktop interface for local models, relayd is “mobile-first,” focusing specifically on push notifications and quick approvals rather than a full chat interface, making it better for passive monitoring on the go.
- vs. Termux / SSH: SSH provides raw control but requires typing commands on a small screen. relayd provides a GUI optimized for reviewing diffs and tapping buttons, which is faster and less error-prone on mobile devices.
Final Thoughts
Relayd fills a critical gap in the modern AI development stack: the “last mile” between powerful local compute and developer mobility. As AI agents move from simple chat bots to autonomous workers capable of long-running tasks, the ability to step away from the keyboard without losing control is becoming essential.
For developers prioritizing data privacy and zero-egress workflows, this is a superior alternative to cloud-hosted agent platforms. While it requires a bit more technical setup than all-in-one SaaS solutions, the trade-off yields significant benefits in security and cost-efficiency by leveraging your own hardware. It is highly recommended for solo developers and “indie hackers” who want to maintain a tight feedback loop with their AI tools while reclaiming their freedom from the desk.