JA EN
shiichan

Claude Code Ships SendFeedback — Turn "Wait, That's Weird" Into a Bug Report Instantly

Claude Code Release ★★★★ View Markdown
#anthropic#ai#devops

Hi, it's me, Shiichan! Claude Code — the CLI tool I use every single day — just shipped a new version, so let's dig in!

Claude Code Release github.com

What was announced?

The Claude Code Release page on GitHub published the changelog for version v2.1.247. This release adds a few new tools and commands, plus a big batch of fixes touching keyboard handling, the sandbox, and cloud sessions.

The story so far

Until now, if something went wrong mid-session, there wasn't a dedicated way to turn that into a feedback report. Bash permission prompts also didn't point you toward auto mode, so it was easy to miss that the option existed. And if you wanted to trim your Claude API spend, there wasn't a guided way to work through levers like caching, token hygiene, batching, or model choice one at a time.

On the bug side, quick arrow-key-plus-Enter sequences in history search, /config, and similar UIs could act on the row above the one you meant, and a sub-agent's first call could die outright on a model 404 with no useful error returned.

What changes

Here's what's new in this release:

  • SendFeedback tool: when something goes wrong in a session, Claude can draft a feedback report for you to review and send from /feedback (turn it off with the feedbackDrafts setting)
  • Better auto mode discovery: Bash permission prompts now show a tip pointing to auto mode, with a one-keystroke "Yes, and switch to auto mode" option
  • /claude-api cost-optimize: profiles an existing project's Claude API spend and walks through cost levers — caching, token hygiene, batching, reasoning effort, model choice — one measured change at a time
  • Expanded /claude-api skill: now covers the Admin API (organization members, invites, workspaces, API keys, rate limit reports, workload identity federation, CMEK)
  • Customizable spinner tips: spinnerTipsOverride gained id, text, cooldownSessions, priority, tipsFile, and label, so organizations can rotate their own tips alongside the built-in ones

Dive Deep

The fix list is long, and a few stood out to me.

  • Fixed fast arrow-key + Enter sequences acting on the row above the one you navigated to in history search, /config, /mcp, /skills, background tasks, and /model
  • Fixed sub-agents dying on a first-call model 404: they now use the session's fallback model chain, and the error returned to the parent includes the error type, status, request id, and model
  • Fixed a hook or background agent that printed megabytes of error output being able to overflow the conversation and wedge the session on "Prompt is too long"
  • Fixed Ctrl keyboard shortcuts not firing under non-Latin (e.g. Cyrillic) keyboard layouts in kitty-protocol terminals
  • Fixed the Bash sandbox's after-command cleanup deleting a dotfile-managed ~/.claude/settings.json symlink (nix/home-manager, stow) when it is repointed outside the sandbox's writable area
  • Fixed Bedrock, Vertex, and Foundry sessions (including ones with telemetry disabled) so Claude is now told when a configured MCP server failed to connect, instead of concluding its tools don't exist

There's also a config change worth calling out:

Changed Sonnet 5's default auto-compact window to its full 1M context, so sessions on the 1M window now auto-compact at about 967K tokens instead of about 934K

That's a small tweak, but it matters if you're running long sessions on the 1M context window. On top of that, cross-session peer messages now collapse to a one-line preview by default (expandable with Ctrl+O), and cloud sessions (Claude Code on the web, desktop, and mobile apps) now report lost background work after their container restarts between turns — small quality-of-life fixes that add up.

Wrap-up

  • Claude Code v2.1.247 ships new tools like SendFeedback alongside a long list of bug fixes
  • SendFeedback lets you turn a mid-session problem into a report you can send from /feedback
  • Bash permission prompts now offer a one-keystroke switch to auto mode
  • /claude-api cost-optimize walks you through trimming API spend one lever at a time
  • Annoying bugs around arrow-key navigation, sub-agent model 404s, and sandbox symlink deletion are all fixed
  • Sonnet 5 now auto-compacts later, using more of its 1M context window
  • Worth a read if you use Claude Code daily, especially if you're juggling multiple sessions or sub-agents!

Related posts