claudekit / updates / claude-code-2-1-293
[ PATCH · ]

Claude Code 2.1.293

Adds Claude Haiku 5.5 (`claude-haiku-5-5`), now the default Haiku model on the Anthropic API. Fixes Claude treating its own last actions before a context compaction as done after it, and path-scoped rules and nested CLAUDE.md files not loading when Claude views a file through the Bash tool. Reverts the 2.1.290 cloud session wakeup fix.

Official announcement →

This article is a summary based on official documentation.

What changed

Claude Code 2.1.293 shipped on October 7, 2026. The headline is Claude Haiku 5.5, now the default Haiku model on the Anthropic API. The rest is mostly reliability work: how Claude reads its own history after a compaction, moving a session to the background with ←, what sub-agents are told about their tools, and several CLI commands that could sign you out. Two earlier fixes are reverted. The items below are a selection of the changes with the most user impact.

New features

  • Claude Haiku 5.5 (claude-haiku-5-5)

    Now the default Haiku model on the Anthropic API, with 1M context at $0.10/$0.50 per Mtok ($0.50/$2.50 for prompts over 100K).

  • agentType in the subagentStatusLine payload

    Scripts can now tell custom subagent types apart.

  • isDeferred on $.tool.register for mods

    A mod’s tool used to sit behind tool search. isDeferred: false lists the tool’s schema in the prompt from the start.

Key improvements

  • Startup for Team and Enterprise organizations

    Policy and managed settings are fetched earlier, and a stalled request is retried after 3 seconds.

  • Claude in Chrome

    Fewer page actions are refused when the browser is slow to report its tabs. In cloud sessions, when the browser can’t be reached and you belong to more than one organization, the message now says the extension must be signed in to the same one, and how to change that.

  • Bash edit diff note

    The note now says the listed files changed while the command ran, which can include writes by other processes.

  • Artifacts

    Claude pins libraries to exact versions two weeks old or older.

Changed behavior

  • claude.ai skill syncing

    Checks for changes about every 40 minutes, instead of every 10, while no session is in use.

  • Sort order of agent lists and MCP servers announced to the model

    Names with non-ASCII characters now sort after ASCII names.

  • OpenTelemetry claude_code.at_mention logging

    Emits at most 100 agent and 100 MCP-resource events each time a prompt is read.

  • Self-hosted runner polling

    The orchestrator sleeps 4 to 6 seconds between polls instead of a constant 5, so several replicas for one environment stop polling at the same moment.

  • Reverted: the 2.1.281 auto mode denial message change

    The message that told Claude a denial covers the outcome, not only the exact command, is gone.

  • Reverted: the 2.1.290 fix for cloud sessions staying asleep

    When a container restart loses a pending /loop wakeup or scheduled task, Claude is no longer told, and the session stays asleep.

Bug fixes

Compaction & MCP

  • Claude sometimes treating its own last actions before a context compaction as done after it, and retracting or redoing finished work.
  • A memory leak where an HTTP MCP connection kept every request it had sent until it closed.

CLAUDE.md & skills

  • Path-scoped rules and nested CLAUDE.md files not loading when Claude views a file with a single-file cat, head, tail, sed -n or grep command in the Bash tool instead of the Read tool.
  • A claude.ai-synced skill’s edited description sometimes not reaching the model until a new conversation or /clear.

Moving sessions to the background (←)

  • A message sent while Claude was working being lost when ← moved the session to the background; if a queued message can’t move, ← now stays put and says so.
  • ← backgrounding a session after 10 seconds while the prompt held unsent text (it now cancels the move) or a question was waiting for your answer.
  • Esc, or No without feedback, on a permission prompt not stopping the turn when ← had just been pressed.

Subagents & tool lists

  • Claude being told to continue or message subagents with SendMessage in sessions, including resumed ones, where a host, a permission rule or a --tools list removed that tool.
  • Subagents and --agent sessions being told a built-in tool was disabled for the whole session when only their own tool list left it out.
  • A custom agent named worker being shown as “Agent” when it starts, and the agent detail dialog’s title losing the agent type once the agent finishes.
  • The footer’s agents count disappearing, and the agents view briefly showing placeholder rows, when the sessions folder could not be read (for example, too many open files).

CLI commands & sign-in

  • claude logs, stop, kill, rm and claude daemon status, stop, uninstall sometimes signing you out when your login had expired or was about to.
  • claude purge stopping silently (exit 0, or a hang in a terminal) when a file or folder could not be deleted; it now deletes the rest, lists what it could not delete, and exits 1.
  • claude agents offering bypass permissions that background sessions then ignored when consent was saved only in .claude/settings.local.json or a --settings file; it now asks for consent first, and a session that ignores bypass shows a short notice that stays.
  • Windows: stopping a status line, hook, or shell command sometimes terminating an unrelated process that had been given the same process ID.

Input & display

  • Pasted text that begins and ends with the same words sometimes being sent to Claude as if it had been typed.
  • A skill name in pasted text being treated as typed when an accent typed or pasted right after the paste merged into its last letter.
  • /model effort ←/→ wrapping past the highest or lowest level, which could accidentally save Low as a model’s default effort.
  • /feedback returning to the drafts list after Ctrl+O or Ctrl+Z while a report was sending, leaving the send impossible to cancel.
  • keybindings.json checks: a lone ” ” (space key) is no longer reported as an error, and keys like “ctrl+ k” now get a warning.
  • Vim mode >> and << on a line of only spaces leaving the cursor past the end of the line, and the cursor position and . after deleting whole lines in Visual mode (V then d).

Plugins & mods

  • A mod’s hooks on classic.* events being skipped while the plugin hooks worker restarts, which left settings hooks to answer without them.
  • claude plugin test failing for mods that call $.session.append; tests can read the appended rows back with the new mock.session.
  • claude plugin eval refusing every Bash-granting run on Macs with Docker Desktop (links under ~/.docker/bin); the refusal now names which part of a credential store held the link.

Remote Control, cloud & integrations

  • Replies in very long Remote Control and cloud sessions appearing a block at a time instead of streaming in.
  • Remote Control uploading a session’s starting history again after every credential recovery.
  • PushNotification reporting “Remote Control inactive” in sessions started with claude remote-control.
  • /tui disconnecting Claude in Chrome in a session started with --chrome, and ignoring --no-chrome.
  • The Claude apps gateway refusing to start when a desktop policy sets Claude Desktop’s built-in browser keys, such as builtinBrowserEnabled.
  • The Artifact tool’s transcript row briefly reading Artifact("(unprintable path)") while a publish call was still streaming in.
  • The /ultrareview upload on Linux wrongly refusing some repositories, such as one inside another checkout, over a settings file that “could not be parsed” while a sandboxed command was running.
  • The /ultrareview upload’s refusal over split-index files advising a git command that could leave git unable to read its index.

Claude Tag & Code Review

  • [Claude Tag] Claude in Slack saying a workspace isn’t set up in Enterprise Grid channels shared across workspaces when Slack labels a message with a workspace that isn’t connected.
  • [Claude Tag] Claude in Slack stopping mid-task when an admin changed the channel’s connectors, plugins, skills or rules; the change now applies once Claude finishes.
  • [Claude Tag] Running a thread’s routine now with extra notes starting a separate run that couldn’t post back; the run now continues in that thread.
  • [Claude Tag] An access bundle’s Add a connector dialog showing every Google connector as connected after one Google sign-in.
  • [Claude Tag] Improvements and changes: admin settings list an Enterprise Grid that isn’t connected yet, with a Connect button; Claude in Slack joins announcement channels without posting its intro; the limit on channel rules goes from 20 to 50 for each workspace and for the organization-wide Slack page.
  • [Code Review] The Add a repository dialog now lists each repository that couldn’t be added and why, such as missing GitHub write access.

Notes

  • If you use Haiku on the Anthropic API, the default is now Claude Haiku 5.5 (claude-haiku-5-5). Prompts over 100K are billed at $0.50/$2.50 per Mtok.
  • If you rely on nested CLAUDE.md files or path-scoped rules, they now load when Claude views a file with cat, grep and similar Bash commands, not only with the Read tool.
  • If you use /loop or scheduled tasks in cloud sessions, the 2.1.290 wakeup fix is reverted, so a session can stay asleep after a container restart loses a pending wakeup.
  • If your agent or MCP server names use non-ASCII characters, they now sort after ASCII names in the lists announced to the model.
  • Mod authors can use isDeferred: false on $.tool.register to list a tool up front, and mock.session in claude plugin test.