Claude Code v2.1.301 brings vim's % motion and richer OTel events!

Hi, it's Shii-chan! I read through the Claude Code v2.1.301 release notes.

What was announced?

The Claude Code Release on GitHub published v2.1.301. It packs in lots of additions, improvements, and fixes.

Here are six additions worth knowing:

CLAUDE_CODE_RETRY_WATCHDOG_MAX_USAGE_LIMIT_WAIT_MS: in unattended retry mode (CLAUDE_CODE_RETRY_WATCHDOG), a request now ends at a usage limit that resets too far away instead of waiting
OpenTelemetry tracing guidance in the claude-api skill, including setting up trace destinations with the ant CLI
The % motion in vim mode: jump to the matching bracket, alone or with d/c/y
rows on a mod's Input: a multi-line text field where Enter submits and Shift+Enter or a backslash before Enter starts a new line
classifier_verdict and permission_mode on OpenTelemetry tool_decision events, so auto mode classifier decisions can be told apart from permission rules
classifier_category on the same events, naming the built-in rule a classifier block matched, when OTEL_LOG_TOOL_DETAILS=1

Some notable fixes and improvements:

Quitting no longer removes a worktree made by a WorktreeCreate hook without asking when it may hold unpushed commits
The fable (and best) alias no longer errors when the newest Fable model isn't available to your organization; Claude Code retries once on the previous Fable model
A full disk that blocks login refresh now gets an error naming the disk instead of a misleading "Please run /login"
Grep and Glob now return an error when the search program fails unexpectedly, instead of reporting no matches
The Read tool no longer reads binary files such as .xlsm, .parquet, .epub, and .heic as text
Switching to a model with a smaller context window now summarizes a too-long conversation with the previous model first, and no longer spends credits on that summary without asking
Conversations are no longer compacted repeatedly behind an LLM gateway that names a different model than requested
Rendering performance improved during long-running subagents
Skills and commands with very many or very long argument names no longer stall (only the first 64 names, each up to 64 characters, are filled in)

Platform-specific fixes:

VS Code: added LaTeX math rendering ($$, \[ … \], and inline math); fixed slowdowns during long thinking and [object Object] in diagnostics
Windows: stopped background PowerShell commands now include the closing [killed] line in their output file
Self-hosted runner: Slack-started sessions can find and attach a repository when managed settings restrict MCP servers (use deniedMcpServers to keep those tools off)

A good update if you run Claude Code unattended or monitor it with OpenTelemetry!