LegalWork v0.0.10

LegalWork v0.0.10

Local model providers

• New "Local model" option under Settings → AI → Connect providers, with one-click templates for the common local runtimes: Ollama, LM Studio, llama.cpp, vLLM, and LocalAI. Pick a runtime — it prefills the endpoint — then fetch your models and connect. No API key needed.
• A locally-running Ollama or LM Studio is still detected automatically and its models appear live; the templates are primarily for reaching a remote or non-default host.

Fixes

• Disconnecting a provider now actually works. Previously, disconnecting a custom or local provider did nothing because only its stored API key was cleared, leaving the provider defined. It's now fully removed, with a clear in-progress state and success/error feedback. (Providers supplied via an environment variable still can't be unset and now say so instead of silently doing nothing.)