kdeps-commands-model

version: 1.0.0
name: model
order: 40
# builtin: kdeps handles this command itself; this file supplies its help,
# aliases and completion.
builtin: true
usage:
  - args: "[name]"
    help: Show or set the LLM model
  - args: "default [name]"
    help: Show or save the default startup model
  - args: list
    help: List all available models with provider status
  - args: ps
    help: List running local model servers (llamafile/gguf)
  - args: "ps kill <model>"
    help: Kill a running local model server
  - args: "ps switch <model>"
    help: Switch to a running local model server
  - args: "hff search <query>"
    help: Search HuggingFace for GGUF models
  - args: "hff info <repo>"
    help: List GGUF files available in a HuggingFace repo
  - args: "hff download <repo> [file]"
    help: Download a GGUF file from HuggingFace
  - args: "<url>"
    help: Register a .gguf/.llamafile URL or OpenAI-compatible endpoint
  - args: "favorite <name>"
    help: "Star a model (shown first in /model, persists); unfavorite to remove"
  - args: "name [show|hide|abbreviate|auto]"
    help: Show or set how the modeline displays the model name
  - args: "tool [list]"
    help: "Show agent loop settings (tool rounds, retries, compaction)"
  - args: "tool set <setting> <value>"
    help: "Change a setting for this session (e.g. rounds 80, retry-delay 5s)"