Feature: Option to restrict the /advisor command to specific models (e.g. disable on Opus)

[AriYacobson]

Summary

Add a configuration option to disable (or skip) the advisor when using specific models, such as Opus.

Motivation

Claude Code has an advisorModel field in settings.json that pairs an advisor model with the executor. When running on Sonnet, this makes sense — the advisor (Opus) provides a quality lift. But when I switch to Opus as my main model, the advisor is still active even though I'm already on the most capable model.

I tested this: with model: "sonnet" and advisorModel: "claude-opus-4-8" configured, switching to Opus via /model claude-opus-4-8 still invokes the advisor. The advisorModel setting is global, not executor-scoped — there is no way to conditionally disable it based on the active model.

Desired Behavior

A way to scope advisorModel so it only applies when a specific executor is active. For example:

{
  "model": "sonnet",
  "advisorModel": "claude-opus-4-8"
}

This should naturally mean: advisor is only active when Sonnet is the executor. When I switch to Opus, no advisor should be invoked.

Alternatively, an explicit disabledForModels option:

{
  "advisorModel": "claude-opus-4-8",
  "advisorDisabledForModels": ["claude-opus-*"]
}

Current Behavior

advisorModel is a flat global setting. Switching the executor model at runtime (via /model) has no effect on whether the advisor is invoked — it remains active regardless of what model is running.

Related

• Add separate effortLevel configuration for advisor model #50502 — requests separate advisorEffortLevel config (different ask, but same area)

[github-actions]

Found 3 possible duplicate issues:

1. Feature request: expose advisor tool as agent/subagent frontmatter field #45964
2. Add separate effortLevel configuration for advisor model #50502
3. [FEATURE] Feature Request: Model-conditional configuration settings (e.g., thinking mode defaults per model) #12388

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

[AriYacobson]

This is not a duplicate of #50502.

• Add separate effortLevel configuration for advisor model #50502 asks for a separate advisorEffortLevel config to control the effort level of the advisor model independently from the executor. It assumes the advisor is already active and focuses on tuning its behavior.

• This issue asks for the advisor to be conditionally disabled based on the active executor model. The problem is that advisorModel is currently a flat global setting — switching the executor (e.g., from Sonnet to Opus via /model) has no effect on whether the advisor is invoked.

I tested this directly: with model: "sonnet" and advisorModel: "claude-opus-4-8" configured, switching to Opus mid-session still triggered the advisor. The two issues are in the same area but request fundamentally different things.

[github-actions]

Closing for now — inactive for too long. Please open a new issue if this is still relevant.