Source Intelligence
Follow Discord

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

← All releases Home ←olderv2.1.9 v2.1.11newer→

Claude Code v2.1.10

7 entries read diff v2.1.9 → v2.1.10 Markdown

This release introduces a new "Push to Remote" option in Plan Mode, allowing users to send their implementation plans directly to a Claude.ai remote session for autonomous execution. It also adds support for hostPattern matching in enterprise marketplace configurations, replaces the command path detection mechanism with a more efficient which-based approach, and lays groundwork for Perfetto performance tracing (currently disabled).

Find
Pick an entry · j / k steps through
6 entries

Changesopen

Push to Remote from Plan Mode#

What

Send an implementation plan directly to a cloud-based Claude.ai session that will execute it autonomously.

Usage: When Claude presents a plan for approval, there's now an option to push the plan to a remote session instead of executing locally. This creates a new remote agent session on Claude.ai that will implement the plan in the cloud.

Details
  • Requires being logged in to Claude.ai (not Console)
  • Requires the Claude GitHub app to be installed on the repository
  • Validates prerequisites before attempting to create the remote session
  • Automatically handles uncommitted or unpushed changes via a git dialog
  • Creates a remote agent that appears in your session list with an r prefix (e.g., r123456)
Evidence

YA() at line 510466 (plan-to-remote handler, creates session with e2A() and tracks with tengu_plan_exit outcome "yes-push-to-remote")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Host Pattern Matching for Enterprise Marketplace Sources#

What

Enterprises can now use regex patterns to allow plugin marketplaces from specific domains without listing each one individually.

Usage: In policySettings.strictKnownMarketplaces, add an entry with source: "hostPattern":

{
  "source": "hostPattern",
  "hostPattern": "^github\\.mycompany\\.com$"
}
Details
  • Matches the hostname extracted from any marketplace source type
  • For GitHub sources, matches against "github.com"
  • For Git sources (SSH or HTTPS), extracts the hostname from the URL
  • Allows enterprise IT to permit all marketplaces from internal GitHub Enterprise instances
Evidence

Yn4() at line 91027 (host pattern regex matcher, uses new RegExp(Q.hostPattern).test(B))

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

ExitPlanMode Tool Parameters for Remote Sessions#

What

The ExitPlanMode tool now accepts parameters to indicate the plan was pushed to a remote session.

Details: New optional parameters:

  • pushToRemote: Boolean indicating the plan was pushed to remote
  • remoteSessionId: The ID of the created remote session
  • remoteSessionUrl: URL to access the remote session
Evidence

Schema at line 416146 (contains pushToRemote, remoteSessionId, remoteSessionUrl parameters)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Improved Command Path Detection#

What

Claude Code now uses native which (Unix) or where.exe (Windows) commands instead of the previous findActualExecutable approach for detecting executable paths.

Details
  • More efficient and reliable command detection
  • Uses Bun.which() when running under Bun runtime
  • Falls back to shell-based which/where.exe on other runtimes
  • Includes both async (GC()) and sync (d1Q()) variants
Evidence

W94() at line 27476 (new which-based implementation, calls where.exe on Windows, which on Unix)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Reduced React DevTools Bundle#

What

React DevTools code has been significantly reduced in the bundle.

Details: References to React DevTools dropped from 45 to 2 occurrences, removing unused debugging infrastructure.

Evidence

Removed variable b6B at line 173391 in v2.1.9 (contained __REACT_DEVTOOLS_COMPONENT_FILTERS__ setup)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Session Naming for Autonomous Tasks#

What

Sessions initiated via autonomous tick-based inputs now display "Autonomous session" as the default title.

Details: When the first prompt starts with <tick>, the session is labeled "Autonomous session" rather than showing the prompt content.

Evidence

ad() at line 5350 (session title generator, contains B ? "Autonomous session" : "" when firstPrompt?.startsWith("<tick>"))

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

2 entries

Bug Fixesopen

#

  • Fixed plugin hooks skipping when allowManagedHooksOnly is enabled in policy settings (Nh() at line 194862, WW2() at line 314806)

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Perfetto Performance Tracing [In Development]#

What

Infrastructure for generating Perfetto-compatible trace files to analyze Claude Code performance.

Status

Feature-flagged, returns false. The PERFETTO_TRACING feature gate is disabled.

Details
  • Set CLAUDE_CODE_PERFETTO_TRACE=1 or provide a file path to enable (once feature flag is enabled)
  • Traces include API call timing, time-to-first-token (TTFT), token throughput metrics
  • Output in Chrome Trace Event format (trace-{sessionId}.json)
  • Tracks per-agent timing with parent-child relationships for nested agent calls
  • Includes metadata like input/output token counts, cache hit rates, and errors

Usage (when enabled):

CLAUDE_CODE_PERFETTO_TRACE=1 claude
# or
CLAUDE_CODE_PERFETTO_TRACE=/path/to/trace.json claude
Evidence

r_2() at line 356811 (initialization, checks process.env.CLAUDE_CODE_PERFETTO_TRACE but guarded by ix.PERFETTO_TRACING which returns !1)

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

Verbatim
Official · Anthropic

Anthropic’s official release notes

Published verbatim by Anthropic for v2.1.10. Text is unmodified from the upstream changelog. Everything else on this page came out of the bundle instead, which is why the two lists don't match.

  • Added new Setup hook event that can be triggered via --init, --init-only, or --maintenance CLI flags for repository setup and maintenance operations
  • Added keyboard shortcut 'c' to copy OAuth URL when browser doesn't open automatically during login
  • Fixed a crash when running bash commands containing heredocs with JavaScript template literals like ${index + 1}
  • Improved startup to capture keystrokes typed before the REPL is fully ready
  • Improved file suggestions to show as removable attachments instead of inserting text when accepted
  • [VSCode] Added install count display to plugin listings
  • [VSCode] Added trust warning when installing plugins
Read it on GitHub →
System prompt

1 of 17 tool descriptions changed.

Agent SDK, print mode

Read the prompts for this release →