Features
Everything a modern reflector needs — nothing cryptic by default
Built for speed and plain English. Underlined terms link to glossary definitions.
All-mode reflecting
D-STAR, Fusion/YSF, DMR, M17, and AllStar today — with P25, NXDN, FreeDV, and EchoLink planned. One reflector, many modes.
Modules with clear purposes
A Main Room, B Chat, C Nets, D Regional, E Tech & Testing, Z Echo Test. Each module has a job so newcomers are not guessing.
All-Rust core (tokio async)
A single static binary or Docker image — memory-safe, fast startup, small footprint. Runs on a cheap droplet or a Raspberry Pi.
Next.js dashboard & browser listening
Live updates and in-browser listening in plain English — see who is where and hear the room without cryptic status dumps.
One-prompt install
Tell your AI coding assistant to install from github.com/adamehrheart/OmniRef using AGENTS.md / SKILL.md — or clone and follow the classic shell path.
Same-mode without fees
Same-mode works with no AMBE dongle and no proprietary codec license fees.
Opt-in cross-mode
Cross-mode voice conversion only when you attach an AMBE dongle locally or reach one over the network.
JSON Schema config + validate
Configuration checked against a schema. Run
omniref validatebefore you go live — humans and agents both benefit.Admin view & admin API
Operate the reflector from a dedicated admin view and API — status, modules, and configuration without SSH archaeology.
AGENTS.md, SKILL.md, planned MCP
Ships with files coding assistants can follow, plus a planned MCP server so AI can check status and manage modules. See /agents.
Opt-in public directory (planned)
A public directory of OmniRef reflectors is planned — operators choose to be listed. Until then, use the live demo at ref1.
Latest technology
Built for speed
OmniRef is built on current tooling for performance: an all-Rust core with tokio async, a single static binary, a Next.js dashboard with live updates and in-browser listening, and modern open standards. It runs on a ~$6/month droplet or a Raspberry Pi — ref1.omniref.io lives on about 1 GB of RAM.
Low latency voice
An async Rust core (tokio) keeps the reflector lean so audio paths stay snappy — built for live radio, not bloated middleware.
Handles many users
One process, modern concurrency, and clear modules. Clubs and public rooms share the same fast path without a stack of mystery services.
Fast startup
A single static binary (or Docker) comes up quickly. Validate config, start the service, and you are on the air.
Tiny footprint
Small memory use so it fits a cheap VPS or a Raspberry Pi. The live demo at ref1 runs on a ~1 GB droplet.
One-prompt install
Install with one prompt to your AI coding assistant
Paste this into Cursor, Claude Code, or any agent that can read AGENTS.md and SKILL.md in github.com/adamehrheart/OmniRef. Replace <IP> and <name> — the assistant can install and configure end to end.
Install OmniRef on my server at <IP> for my club <name> using github.com/adamehrheart/OmniRef
git clone https://github.com/adamehrheart/OmniRef.git && cd OmniRef && cat AGENTS.md
Prefer the AI prompt above. The shell line clones the repo so you can follow AGENTS.md yourself.
Default modules
Rooms named by purpose — the opposite of cryptic talkgroup folklore.
Module A
Main Room
The everyday hangout — general QSOs and meeting new operators.
Module B
Chat
Casual conversation when you want a quieter, talkative room.
Module C
Nets
Scheduled nets and structured check-ins.
Module D
Regional
Local and regional traffic for clubs and nearby operators.
Module E
Tech & Testing
Experiments, new modes, and troubleshooting without cluttering the main room.
Module Z
Echo Test
Hear yourself back so you can check audio before you go live.