Claude Code v2.1.232: Subagent Forking, GitLab Support, and Security Fixes

Claude Code v2.1.232 is out, and it brings a few significant changes: subagent forking is now default, GitLab support lands in plugin marketplaces, and several security patches address permission bypasses. If you're using Claude Code for multi-session workflows or in enterprise environments, this update is worth a look.
Subagent Forking On by Default
The headline feature: subagent_type: "fork" is now enabled by default. Forked subagents inherit the full conversation and prompt cache, which means they start with all the context you've built. Also, non-teammate agent spawns in interactive sessions now run in the background by default — no more blocking the main session.
Cross-Session Messaging via @
You can now type @ in the prompt to mention another Claude session by name. Claude uses SendMessage to reach that session directly. Bare names that exactly match a live session are delivered immediately (no confirmation prompt with a ref). Session names are also enforced to be unique on a machine; if you try to start or rename to a name already in use, you'll get a name-word-word variant and a notification.
New /config rows let you control "Dialog expiry" and "Messages from your other sessions" — the latter offers accept/hold/refuse for cross-session inbound requests.
GitLab Everywhere
Secret redaction now covers a whole family of GitLab tokens: glrt-, gloas-, glptt-, glagent-, glimt-, glsoat-, glcbt-, glft-, glffct-, plus full redaction of routable glpat-/gldt- tokens. The glab CLI config store gets the same sandbox and credential-path protection as gh.
Plugin marketplaces now support GitLab: bare gitlab.com repo URLs (including nested subgroups) clone just like GitHub URLs, and clone auth-failure hints name your actual Git host.
Settings and Enterprise Policy
additionalMarketplacesandallowedMarketplacesare accepted as friendlier aliases forextraKnownMarketplacesandstrictKnownMarketplaces.- A URL-typed
blockedMarketplacesentry for a bare repo URL continues to block that URL when the CLI classifies it as a Git clone. - Gateway: the
desktop:overlay now accepts every released Desktop setting (was 11 hand-listed keys), validated at boot against Desktop's own schema. Unknown or invalid keys fail boot. - Gateway: empty
managed.policies[].match.groups/admin.admin_groupsentries and malformedemail_domainvalues (empty, or containing@, whitespace, or commas) now fail at boot instead of silently matching no one or granting admin access.
Security Fixes
- Fixed a PowerShell permission bypass where variable-writing parameters could silently overwrite
$PSDefaultParameterValuesand redirect later commands' file access. - Fixed a Windows permission bypass where Git Bash followed Cygwin-style symlinks that path validation saw as regular files; writes through them now require permission approval.
- Fixed nested git repositories inheriting trust from a parent directory; each repository now requires its own trust confirmation.
Other Notable Fixes
- MCP connections no longer hang for the full 30-second connect timeout when a server fails to answer or sends a malformed reply to the protocol-version probe.
- Remote Control: sessions hosted by a bridge inside a cloud session no longer inherit that session's transcript or credentials; sessions started from Desktop/IDE reattach to the existing claude.ai session; idle sessions are reachable; bridge sessions restore conversation history on worker restart; resuming a deleted session starts a replacement instead of failing.
- Cloud gateway
/loginshows the reason when managed settings fail to load, instead of exiting silently or leaving an unresponsive terminal. - Voice mode on native builds shows rejection immediately when the voice service rejects the connection.
- mTLS client certificate rotation no longer requires a restart; Claude Code reloads the rotated cert and key automatically on connection errors.
- Malformed AWS or Vertex region values fall back to the default region instead of being used to build request URLs.
- Stream idle timeout errors recover properly on Bedrock, Vertex, and gateway deployments.
- Overlay rendering fixed for content-sized overlays with truncated text.
Also, Fable 5 is back as an advisor in /advisor for organizations with Fable access, with usage-credits consent set up through /model fable.
If you're running Claude Code in enterprise or multi-session setups, this release tightens security and improves cross-session interactions. Update and check your configurations.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Qwen 3.6 27B Benchmarked on DeepSWE: 2% Score, 70 Hours, 44k Avg Output Tokens
Qwen 3.6 27B (FP8, BF16 KV cache, 262k context) scored 2% on DeepSWE in 70 hours. Output tokens averaged 44k per task — comparable to larger models like Qwen 3.6 Plus. Ran on 1x RTX6000 Pro Blackwell via RunPod.

VS Code to Enable Co-Authored-by Copilot Trailer by Default
Microsoft's VS Code PR #310226 changes the git.addAICoAuthor setting default from 'off' to 'all', automatically adding a Co-authored-by trailer for AI-generated contributions. The PR also reveals a runtime fallback mismatch in repository.ts.

Claude Opus 4.7 System Prompt Changes: Platform Renaming, Tool Integration, and Behavioral Updates
Anthropic updated the Claude Opus system prompt from version 4.6 (February 5, 2026) to 4.7 (April 16, 2026), renaming the 'developer platform' to 'Claude Platform', adding Claude in Powerpoint to the tools list, expanding child safety instructions, and implementing new behavioral guidelines for tool usage and response conciseness.

Claude Code Auto Mode Becomes Default: Why Anthropic Stopped Trusting Humans
Claude Code's Auto Mode, which runs agentic tasks with minimal human intervention, is becoming the default due to a belief that human oversight slows workflows and introduces errors.