
ShunyaNet Sentinel is a lightweight program that monitors RSS feeds, sends them to an LLM for analysis, and delivers summaries or alerts directly to the GUI and Slack.
ShunyaNet Sentinel is a lightweight, cyberpunk-themed program that ingests RSS feeds (breaking news, social media, government alerts), sends them to an LLM for analysis, and delivers alerts and summary reports to the GUI and your notification channels at regular intervals.
The project uses an LLM hosted locally on the same machine or network (e.g., with Tailscale) via LM Studio, Ollama, or llama.cpp — or the OpenAI API. Alerting works over Slack, Discord, Telegram, ntfy, any generic webhook, and desktop popups (use as many at once as you want).
Compatible with current versions of Linux, macOS, and Windows.
The quality of reporting depends on your prompt, context size, RSS feeds, and LLM. You may need to turn off "thinking" features if your model has them. I use this most with Qwen3.8 27B and Qwen3.8 Flash Next. Gemma models should be pretty good, too.
agent_handoff.json index so scripts and AI agents can consume Sentinel's output without touching the GUI/v1/chat/completions endpoint: LM Studio, Ollama, llama.cpp, OpenAI#> ALERT / #> SUMMARY / #> MEMORY sectionsOther than polling RSS feeds and sending your notifications (both optional), all information stays on hardware you control.
Sentinel writes stable, machine-readable artifacts beside every report: Report_Output/agent_handoff.json points at the newest alert/summary file plus the memory files. Any local agent can poll that one JSON file to follow what Sentinel is seeing. See docs/AGENT-INTEGRATION.md for the full contract, and Help → Agent / Script Integration inside the app.
requirements.txt): PySide6, feedparser, requestsLLM hosted via:
git clone https://github.com/EverythingsComputer/ShunyaNet-Sentinel.git
cd ShunyaNet-Sentinel
macOS / Linux:
python3 -m venv venv
source venv/bin/activate
Windows (PowerShell):
python -m venv venv
venv\Scripts\Activate.ps1
pip install -r requirements.txt
macOS / Linux: python3 ShunyaNet_Sentinel.py
Windows: python ShunyaNet_Sentinel.py
On first launch, Sentinel walks you through setup with a guided Setup Wizard (reopen anytime with Ctrl+W or Help → Setup Wizard):
Manual setup (skipping the wizard):
.txt, one RSS URL per line — your loaded list is remembered across restarts).Ctrl+,) → Connection tab: choose provider, set the Server URL (LM Studio example: http://localhost:1234/v1), press Test Connection, Save & Apply.Ctrl+Enter, or wait for the auto-fetch countdown in the status bar)To push alerts to your phone: Settings → Notifications → fill one channel (Slack/Discord/Telegram/ntfy), press Send Test Notification, Save & Apply. Setup hints for each service are in that tab's tooltips.
Every setting has a hover tooltip; Help (F2) lists keyboard shortcuts. Matrix-rain screensaver: F1.
Topics — enter topics in the Targeting panel (autosaves). Save named sets as Profiles to switch watch-lists quickly. Your loaded feed list is part of app state and persists across restarts.
Sentinel ships with 11 ready-made topic profiles (topic_profiles.default.json) that are installed automatically on first run into your own topic_profiles.json — after that the file is yours: edit, delete, add, and your changes are never overwritten. Brought a fresh clone or deleted one you liked? Profiles → Restore Bundled Topic Profiles… merges back only the shipped profiles you are missing.
A profile has two fields, matching the Targeting panel: overall_topic (one sentence framing what counts as news) and subtopics (newline-separated signal lines — write each as what to look for, plus examples in parentheses; specific signals screen far better than broad subject words).