Model Router Configure

Configure settings for the Model Router to optimize AI model usage.

Change Model Router settings.

Trigger

User runs /model-router:configure or /model-router:configure [setting]

Arguments

  • setting (optional): Specific setting to change. One of:
    • subscriptions - Update subscription status
    • routing - Change routing preference
    • profile - Change profile
    • tracking - Enable/disable tracking
    • ollama - Configure Ollama
    • cli - Set up CLI tools

If not provided, show menu of options.

Instructions

1. Load Current Configuration

Read ~/.model-router/config.yaml

If doesn't exist:

No configuration found. Run /model-router:setup for initial setup.

2. Show Configuration Menu (if no argument)

What would you like to configure?

1. Subscriptions - Update which subscriptions you have
2. Routing preference - Change how models are prioritized
3. Profile - Switch between quality/balanced/budget/local
4. Tracking - Enable or disable usage tracking
5. Ollama - Configure local models
6. CLI tools - Set up Codex or Gemini CLI
7. Reset - Start fresh with setup wizard

Current settings:
  Profile: [current]
  Routing: [current]
  Tracking: [current]

Enter number or setting name:

3. Handle Each Setting

Subscriptions:

Current subscriptions:
  Claude Pro:    [✓/✗]
  ChatGPT Plus:  [✓/✗]
  Google AI Pro: [✓/✗]

Which would you like to change?
□ Claude Pro ($20/mo)
□ ChatGPT Plus ($20/mo)
□ Google AI Pro ($20/mo)

Update config and offer to set up missing CLI tools.

Routing Preference:

Current: [preserve_claude/quality_first/budget_first/ask_each_time]

Choose routing preference:
○ ChatGPT + Gemini first (preserve Claude)
○ Claude first (best quality)
○ Cheapest first (maximize free/local)
○ Ask me each time

Profile:

Current: [quality/balanced/budget/local_only]

Choose profile:
○ Quality - Always use the best model
○ Balanced - Smart routing based on complexity
○ Budget - Maximize cheap/free models
○ Local-only - Only use Ollama

Tracking:

Current: [disabled/basic/detailed]

Choose tracking level:
○ Disabled - No tracking
○ Basic - Task type, model, success/fail
○ Detailed - Include quality ratings and timing

If enabling, explain:

Tracking helps Model Router learn which models work best for you.
Data is stored locally at ~/.model-router/tracking.jsonl
Run /model-router:learn to analyze patterns.

Ollama:

Current Ollama config:
  Enabled: [yes/no]
  Host: [localhost:11434 or custom]
  Models: [list]

What would you like to do?
○ Enable/disable Ollama
○ Change host address
○ Pull new models
○ Remove models from routing

If pulling models, guide through ollama pull [model].

CLI Tools:

CLI Tool Status:
  Codex: [✓ Ready / ⚠ Not auth / ✗ Missing]
  Gemini: [✓ Ready / ⚠ Not auth / ✗ Missing]

What would you like to do?
○ Set up Codex CLI (for ChatGPT Plus)
○ Set up Gemini CLI (for Google AI Pro)
○ Re-authenticate existing CLI

Walk through installation/auth as in setup skill.

4. Save Configuration

After any change, update ~/.model-router/config.yaml:

# Updated field
preferences:
  routing: [new_value]

Confirm:

Configuration updated.

Changed: [setting] → [new_value]

Run /model-router:status to see full configuration.

5. Reset Option

If user chooses reset:

This will delete your current configuration and run setup again.

Your tracking data (if any) will be preserved.

Continue?
○ Yes, reset configuration
○ No, cancel

If yes, delete config and run /model-router:setup.