LLM Plugin V2.0 for VisualNeo Win

By Christopher Calleja — Kriscall Software Labs Free for the VisualNeo community

Major update to the multi-provider AI plugin. Now at 25 actions across 7 providers (OpenAI, Claude, Google Gemini, Ollama, OpenRouter, Groq, Grok).

What’s New

3 New Actions:

LLMGetEmbedding — Vector embeddings for semantic search/similarity (OpenAI, OpenRouter, Ollama)
LLMSendWithDoc — Send a document file with your prompt for AI analysis (supports 19 text formats up to 100KB)
LLMClearHistory — Clear conversation without reinitializing the provider

Test Connection — New button in the Init dialog. Tests your API key and provider instantly — shows the AI’s response or a clear error message.

Variable Picker (“…”) — All 11 variable fields now have a picker button matching standard VisualNeo workflow.

Redesigned Dialogs — All 21 edit dialogs now have yellow help panels with detailed usage guidance, bold labels, footer with plugin icon, and a consistent professional look. New AI brain logo in the About dialog.

Key Fixes

OpenRouter free models (:free) no longer fail with 402 errors
OpenRouter defaults to free model google/gemma-3-1b-it:free
Google Gemini system prompt now persists across turns
JSON parser handles escaped characters correctly
Sync calls (LLMSend/LLMImageGen) now show “Thinking…” progress
Streaming buffer no longer limited to 64KB
Async thread safety improvements
Proper UTF-8 handling for international characters
15 bug fixes total

Self-Contained

No external dependencies. Single .nbp file — just copy to your PlugIns folder and go. Statically linked, no redistributables needed.

Getting Started

LLMInit “OpenRouter” “[ApiKey]” “[Status]”
LLMSetModel “google/gemma-3-1b-it:free”
LLMSend “Hello from VisualNeo!” “[Answer]”

MIT License.

(c) 2026 Kriscall Software Labs