NodeBrain v0.3.6
Features
- Custom provider endpoints — you can now set a custom Base URL on a provider, so you can point NodeBrain at any OpenAI-compatible endpoint, including an Ollama or other inference server on your LAN, instead of being locked to localhost.
- Optional API key for Ollama — the credential form now shows an optional Base URL field for Ollama instead of an API key field, and saves without a key.
Fixes
- Scheduled agents now use their current model. Previously a scheduled agent could keep running with a stale model after you changed it (e.g. still requesting
llama3.2after switching to another Ollama model); scheduled runs now re-read the agent's current settings at run time. - Custom Ollama model names are now used correctly for requests instead of falling back to a default.
Notes
- Code signing is still pending, so the unsigned Windows build may show a SmartScreen warning (More info → Run anyway).
⚠️ Known issue: v0.3.6 fails to start on a fresh install. Fix coming in v0.3.7 — please wait.