Iris Remote Client

Cross-platform desktop client (Linux, Windows, macOS) for the IrisRemoteServer running inside the macOS Iris menu-bar app. The AI logic stays on the Mac — this client is a thin remote viewer, analogous to the iOS companion app.

Features

• Connect to IrisRemoteServer over HTTPS with Bearer Token auth
• Create agent jobs (Hermes / Claude Code / opencode)
• Stream job output live via SSE
• Send follow-up messages to running jobs
• Browse and re-open previous jobs
• Config persisted per platform (~/.config/iris-remote/ on Linux)

Tech Stack

What	Library
UI	PyQt6
HTTP + SSE	httpx (async, verify=False for self-signed certs)
Qt ↔ asyncio	qasync
Config paths	platformdirs

Setup

# Create venv (once)
python3 -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run
python main.py

Configuration

On first launch, open the Einstellungen tab and enter:

• Server-URL — e.g. https://192.168.1.x:8765 or your Tailscale address
• Bearer Token — from Iris → Settings → Remote Server

Click Verbindung testen to verify, then Speichern. Settings are stored in ~/.config/iris-remote/config.json (never in the repo).

Project Structure

linux-client/
├── main.py                  # QApplication + qasync event loop
├── requirements.txt
├── client/
│   ├── models.py            # AgentJob, IrisConfig dataclasses
│   └── server.py            # IrisServerClient — all HTTP calls + SSE
└── ui/
    ├── main_window.py       # Main window, tab management, client lifecycle
    ├── settings_view.py     # Server URL + token, connection test
    ├── job_list_view.py     # Job list with 5 s polling, new job button
    ├── job_detail_view.py   # Streaming output + follow-up input
    └── new_job_dialog.py    # Agent type picker + prompt input

IrisRemoteServer API

The client talks to IrisRemoteServer (default port 8765) running on the Mac:

Endpoint	Method	Purpose
/health	GET	Connection test + version
/agents	GET	List all jobs
/agent	POST	Create new job
/agent/:id/stream	GET	SSE stream (job output)
/agent/:id/followup	POST	Send follow-up message

SSE format: data: {"chunk":"..."} / data: [DONE] / data: {"error":"..."}

Requirements

• Python 3.12+
• IrisRemoteServer reachable on the network (LAN or Tailscale)