Radar
The full edition, from release notes to field guides.

Local or cloud? Follow the work.
Codex or Hermes decides how the job runs. Qwen or Luna decides where the thinking happens. Here’s how to pair them for a code fix, a private document or a daily digest.

Run your first service with Docker Compose
Serve a page from a folder on your computer. Replace the container and prove the page survives.

Set up a NAS share you can actually restore
Build a small Synology share, check who can write to it, and recover yesterday’s file to a second folder.

Open a local web app privately from your phone
Connect a loopback-only service to an existing tailnet and verify access over cellular.

Bring up an Ubuntu VPS and prove you can recover it
A first-hour walkthrough for a disposable server, from SSH login to a working page and a recovery check.

Try a new MCP server without guessing what your client supports
Use the July 2026 protocol revision and a read-only tool to check a client–server pairing.

Build a regression test for an AI release classifier
Score a prompt on labeled examples before changing models or quantization. The test runner uses only Python.

Make an AI summary earn each citation
Build a small source-grounded writing workflow that can abstain when the source is silent.

Which Qwen3.8 quant belongs on your GPU?
Seven actual GGUF downloads show what GSQ-RCO, IQ3_S, Q3_K_S and ternary packing buy you on a 24 GB card.

Hermes or Open WebUI? Choose by the job.
One can deliver a report while you sleep. The other gives your household a shared place to talk to models. Here’s where each earns its keep.

Hermes ships a patch across its stable channels
Version 0.23.3 brings a simpler CLI setup and changes to how long tool-running sessions manage their context.

Immich fixes two quiet failures in library monitoring
The 3.4.2 patch repairs integrity-report downloads and the Prometheus exporter. Both matter when checking on a photo archive.

Map a Docker app’s data before an upgrade
A short, read-only inventory shows which files survive a container replacement and which still need a backup plan.

A photo backup needs its catalog, too
Original files and the Immich database answer different recovery questions. Test both with a disposable library.

How much of a model is actually in VRAM?
Ollama exposes the loaded model, its quantization and its reported VRAM size in one local request.

Read the number behind the model name
Qwen3-32B has 32.8 billion parameters. That difference matters when estimating a tight memory budget.