Claude Code v1.0.40: Faster /resume, Smoother Terminal Scrolling, and Inline Thinking Progress

In short. Anthropic shipped Claude Code v1.0.40 on May 1, 2026. The headline wins are a 67% speedup on /resume for sessions over 40MB, a /terminal-setup command that tunes scroll sensitivity inside VS Code, Cursor, and Windsurf terminals, and a thinking spinner that now reports its own progress inline.

What Changed

Claude Code v1.0.40 dropped on May 1, 2026. The release notes (CHANGELOG.md) call out five shipping changes worth flagging for daily users:

/resume is up to 67% faster on sessions over 40MB. Large sessions with many dead-fork entries also load more efficiently. If you have ever waited 10+ seconds to pull a long Claude Code conversation back into focus, this is the single most felt change in the release.
/terminal-setup now configures scroll sensitivity inside VS Code, Cursor, and Windsurf terminals. The fullscreen UI used to feel laggy or jumpy when scrolling fast. The command writes the right editor settings so the scroll behavior matches what you would expect from a native terminal app.
The thinking spinner reports progress inline. Instead of a separate hint row, you see “still thinking”, “thinking more”, and “almost done thinking” attached to the spinner itself. It is a small visual change that makes long thinking blocks feel less like a freeze.
PR branch decoration in the footer renders correctly regardless of model name length. Long model identifiers (Sonnet 4.6, Opus 4.7) used to truncate or push the branch label off-screen. Fixed.
/clear and /new reset the active custom agent selection. A small but useful behavior change if you switch between custom agents inside one terminal session.

Why It Matters

If you live in Claude Code for sustained sessions (debugging across hundreds of files, iterating on a refactor over a workday), the /resume speedup is the change you will feel first. Pulling a 50MB transcript back into context used to be the slowest part of resuming a long task. Cutting that wait by two-thirds removes one of the bigger friction points in long-running agent work.

The /terminal-setup improvement matters for anyone who runs Claude Code inside a code editor terminal pane rather than a dedicated terminal app. VS Code, Cursor, and Windsurf all ship default scroll settings tuned for log viewing, not for fullscreen TUIs. The new command writes the correct settings automatically. Run it once per editor.

The inline thinking spinner is cosmetic but useful. Claude Code thinking blocks can take 10 to 60 seconds on hard problems. When the UI looked frozen, the temptation was to hit Ctrl+C and retry. The visible progress indicator removes that ambiguity.

How to Use It

Update Claude Code to v1.0.40 or later:

npm install -g @anthropic-ai/claude-code
claude --version

If you use Claude Code inside a code editor terminal, run /terminal-setup once to apply the scroll settings:

/terminal-setup

If you have custom agents configured, note that /clear and /new now reset the active agent. If you want to preserve the agent selection across a session reset, switch agents explicitly after running /new.

If you are still picking between coding agents, our Claude Code vs Cursor comparison covers how the two stack up on price, model access, and IDE integration. The Cursor vs Copilot guide is the right starting point if you want a side-by-side on the two most-installed coding agents. For broader context on the coding agent landscape, see our best AI coding tools roundup.

If Claude Code’s terminal-first workflow is not your speed, Cursor and GitHub Copilot both run in-editor with native UI rather than in a TUI.

This post is part of Pondero’s daily coverage of AI tool updates. See all coding guides →