shield_lock
JinDaGe
← Back to skills
extension
Category: Content & MediaNo API key required

model-selection-orchestrator

Select the best local Ollama model for a task and emit tuning presets. Use when choosing between reasoning, creative, fast, vision, synthesis, or pre-AGI modes based on available local models.

personAuthor: jakexiaohubgithub
open_in_newRepositorydownloadDownload package

Model Selection Orchestrator

Choose an Ollama model based on task intent and emit a consistent tuning preset.

Workflow

  1. List local models with ollama list.
  2. Map the task to a preference list and pick the first available model.
  3. Emit a recommended model, a backup, and a tuning preset.
  4. Fall back to the fastest available model when none match.

Scripts

  • Run: python skills/tuning/model-selection-orchestrator/scripts/recommend_model.py --task reasoning

References

  • references/presets.json