[FEATURE] Show agent reasoning/thinking activity in Claude Code on the web

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Claude Code on the web (claude.ai/code) shows tool use — file reads, commands, edits — but not the agent's reasoning. During multi-minute sessions running on remote Anthropic-managed VMs, there's no way to tell whether the agent is actively thinking, what plan it's pursuing, or if it's heading down a wrong path. The session can appear frozen even when work is happening.

Proposed Solution

At minimum: a "Thinking..." indicator (ideally with a token counter or elapsed time) so users know the agent is active rather than stuck.

Better: render thinking blocks inline in the session timeline, collapsed by default with a one-line preview (e.g., "Planning OAuth integration..."), expandable on click. Visually distinct from tool use — dimmed/italic, like the terminal.

Alternative Solutions

_No response_

Priority

Critical - Blocking my work

Feature Category

CLI commands and flags

Use Case Example

_No response_

Additional Context

_No response_