torque-config

TORQUE Config

View and modify TORQUE system configuration.

Instructions

If no argument — show full config overview:

1. Call in parallel:

   • configure (no args) — general settings
   • get_llm_tuning — model tuning parameters
   • get_hardware_tuning — GPU/CPU settings
   • check_ollama_health — host status

2. Present as:

## TORQUE Configuration

### General
| Setting | Value |
|---------|-------|
| max_concurrent | X |
| default_timeout | X min |

### LLM Tuning
| Parameter | Value |
|-----------|-------|
| temperature | X |
| top_p | X |
| preset | X |

### Hardware
| Setting | Value |
|---------|-------|
| num_gpu | X |
| keep_alive | X |

### Hosts
[Per-host status and settings]

If argument is a category:

• "tuning": Show get_llm_tuning details. Offer preset selection via AskUserQuestion (code, precise, creative, balanced, fast).
• "hardware": Show get_hardware_tuning. Offer to run benchmark via AskUserQuestion.
• "safeguards": Show current safeguard config. Offer toggles for: quality_scoring, build_check, auto_rollback, validation.

If argument is key=value:

Parse the setting and route to the correct tool:

• max=N or max_concurrent=N → configure
• timeout=N → configure
• temperature=N, top_p=N → set_llm_tuning
• preset=X → apply_llm_preset
• num_gpu=N, keep_alive=X → set_hardware_tuning
• quality_scoring=0|1, build_check=0|1, auto_rollback=0|1 → configure

Apply the setting and confirm the new value.