Claude Code · setup

/advisor: let a stronger Claude check Opus at the moments that matter

Claude Code's advisor tool pairs your main model with a stronger one. The main model does all the work; the advisor reads the whole session and steps in before Claude commits to a plan, when an error keeps coming back, and before it calls a task done. Here's the setup and which pairs work.

Came from the Reel? You commented ADVISOR. Here is everything, with one-tap copy.

Before you start: Experimental, Anthropic API only (not Bedrock, Vertex or Foundry). Works on subscription and API accounts. Fable as advisor needs Fable access and, on some plans, a one-time usage-credits consent via /model fable.

Setup

Pick an advisor from the list
/advisor
Or set it directly (Fable isn't in the picker, pass it by name)
/advisor fable
Turn it off any time
/advisor off

All 1, step by step

Pairs that work

The advisor has to rank at or above your main model. Haiku can call an advisor but can't be one.

Sonnet main + Opus advisor: Sonnet does routine work, Opus handles planning and stubborn failures.
Sonnet or Opus main + Fable advisor: Fable's judgment at decision points without running Fable throughout.
Haiku main + Opus advisor: the cheapest main model with strong planning.
Opus main + Opus advisor: a second Opus reviews the first for high-stakes work.

Command

{
  "advisorModel": "opus"
}

Tip: Want a check right now? Say "consult the advisor before you continue" in your prompt.

Official docs →

Tip: Claude calls the advisor at decision points, not every turn, so a fast main model plus a strong advisor usually costs less than running the strong model the whole time.

/advisor: let a stronger Claude check Opus at the moments that matter

Setup

All 1, step by step

Pairs that work

More from our Reels