Thinking Modes

Overview

Thinking mode controls how much reasoning effort YeePilot requests before producing responses/plans.

Available values:

• off
• on
• low
• medium
• high

In-Session Control

/think off
/think on
/think low
/think medium
/think high

Config Default

ai:
  think_mode: medium

When to Use What

• off: very simple checks and low-latency interactions
• low: routine multi-step tasks
• medium: default for troubleshooting and planning
• high: complex root-cause analysis, migration planning, and high-risk design changes

Interaction with Other Controls

• Use /plan <task> for explicit plan-first workflows.
• Use /autonomy ... to control execution strictness.
• Use /tokenmode saver if higher thinking depth increases token usage too much.