Always Show Thinking

Show thinking block content inline instead of collapsed behind "∴ Thinking (ctrl+o to expand)".

ID always-show-thinking · Default on · Compatible * · Source

What it does

Claude's thinking blocks render in full every turn, the way transcript mode and --verbose already show them. You never see the collapsed "∴ Thinking (ctrl+o to expand)" placeholder.

Why

Claude Code hides thinking by default to save vertical space, but for anyone who actually reads what Claude is reasoning about, the collapse is pure friction — especially while debugging a prompt or a tool call. Tracked upstream at anthropics/claude-code#8477 (195 reactions at time of writing).

How it works

Inside AssistantThinkingMessage, the gate is:

if (!(isTranscriptMode || verbose)) { return <collapsed view> }

The patch finds the function by searching for the ∴ Thinking string (the "therefore" character, U+2234, is unique to this component), walks up to the enclosing function, finds the first IfStatement whose test is !(X || Y), and replaces the test with false. The collapsed branch becomes dead code and the expanded render always runs.