Claude Code v2.1.119: Config Persistence, GitLab/Bitbucket PR Support, and Dozens of Bugfixes

Anthropic just shipped Claude Code v2.1.119, a hefty maintenance release with over two dozen bugfixes and a handful of notable feature additions that tighten the tool for daily use. Here's what's worth your attention.
Config Persistence & Custom PR URLs
/configsettings (theme, editor mode, verbose, etc.) now persist to~/.claude/settings.json. These participate in the project/local/policy override precedence chain, so you can set team-wide defaults and let local overrides win.- New
prUrlTemplatesetting lets you point the footer PR badge at a custom code-review URL instead of github.com — useful for self-hosted GitLab or Bitbucket. - New environment variable
CLAUDE_CODE_HIDE_CWDsuppresses the working directory in the startup logo.
Broader PR Support & Agent Improvements
--from-prnow accepts GitLab merge-request, Bitbucket pull-request, and GitHub Enterprise PR URLs — no more manual URL translation.--printmode now honors the agent'stools:anddisallowedTools:frontmatter, matching interactive-mode behavior.--agent <name>now respects the agent definition'spermissionModefor built-in agents.- PowerShell tool commands can now be auto-approved in permission mode, matching Bash behavior.
Hooks, MCP, and Telemetry
- PostToolUse and PostToolUseFailure hook inputs now include
duration_ms(tool execution time, excluding permission prompts and PreToolUse hooks). - Subagent and SDK MCP server reconnection now runs in parallel instead of serial — noticeable speedup when many servers are reconfigured.
- Plugins pinned by another plugin's version constraint now auto-update to the highest satisfying git tag.
- OpenTelemetry:
tool_resultandtool_decisionevents now includetool_use_id;tool_resultalso includestool_input_size_bytes. - Status line stdin JSON now includes
effort.levelandthinking.enabled.
Bugfixes Worth Noting
- Fixed CRLF content pasting (Windows clipboards, Xcode console) that inserted an extra blank line between every line.
- Fixed multi-line paste losing newlines in terminals using kitty keyboard protocol inside bracketed paste.
- Fixed Glob and Grep tools disappearing on native macOS/Linux builds when Bash tool is denied via permissions.
- Fixed scrolling up in fullscreen mode snapping back to the bottom after every tool finish.
- Fixed MCP HTTP connections failing with "Invalid OAuth error response" when servers returned non-JSON bodies for OAuth discovery.
- Fixed auto mode overriding plan mode with conflicting "Execute immediately" instructions.
- Fixed async PostToolUse hooks with no response payload writing empty entries to the session transcript.
- Fixed
${ENV_VAR}placeholders inheadersfor HTTP/SSE/WebSocket MCP servers not being substituted before requests. - Fixed MCP OAuth client secret stored via
--client-secretnot being sent during token exchange for servers requiringclient_secret_post. - Fixed
@-fileTab completion replacing the entire prompt when used inside a slash command with an absolute path. - Fixed stray
pcharacter at startup in macOS Terminal.app via Docker or SSH. - Fixed
/exportshowing current default model instead of the conversation's actual model. - Fixed verbose output setting not persisting after restart.
- Fixed
/usageprogress bars overlapping with "Resets …" labels. - Fixed
/planand/plan opennot acting on the existing plan when entering plan mode.
Vertex AI: Tool Search Disabled by Default
On Vertex AI, tool search is now disabled by default to avoid an unsupported beta header error. Opt in with ENABLE_TOOL_SEARCH.
Upgrade
Run npm update -g @anthropic-ai/claude-code or check the release notes for full details.
📖 Read the full source: GitHub Claude-Code
👀 See Also
The Mundane Risk: Why AI Safety's Biggest Threats Are Boring, Not Dramatic
An essay argues that mundane AI failures are already causing damage at scale, current alignment approaches depend too heavily on sandboxed environments, and capability convergence makes accidental open-world exposure increasingly plausible.

IDP Leaderboard benchmark shows Claude Sonnet 4.6 matches Opus 4.6 for document AI tasks
The IDP Leaderboard tested 16 AI models on 9,000+ documents across OCR, table extraction, key extraction, visual QA, handwriting, and long documents. Claude Sonnet 4.6 scored 80.8 overall, essentially matching Opus 4.6 at 80.3, while Haiku 4.5 scored 69.6.

Claude Sonnet 4.6 Beats Opus 4.6 on Execution in Prompt Benchmark
A Reddit user submitted a complex prompt to both Sonnet 4.6 and Opus 4.6; the Sonnet model produced a superior response judged by creativity and hidden requirements.

Polsia Platform Shows Repetitive SaaS Patterns in Live Founder Launches
Polsia is an autonomous business platform where users describe their business, pay money, and it executes autonomously. A behavioral scientist observed 72 hours of live founder launches, identifying repetitive patterns like AI SDR automation solutions and underserved international markets.