Claude Code v2.1.143: Plugin Dependency Enforcement, PowerShell Defaults, and Background Session Fixes

Claude Code v2.1.143 is out, bringing plugin dependency management, PowerShell improvements, and a slew of bug fixes targeting background sessions, Windows usability, and macOS file permissions.
Plugin Dependency Enforcement
claude plugin disable now refuses when another enabled plugin depends on the target. It provides a copy-pasteable disable-chain hint. claude plugin enable force-enables transitive dependencies. The /plugin marketplace browse pane now shows projected context cost (per-turn and per-invocation token estimates).
PowerShell Changes
The PowerShell tool now passes -ExecutionPolicy Bypass by default. Opt out with CLAUDE_CODE_POWERSHELL_RESPECT_EXECUTION_POLICY=1. On Windows, it is enabled by default for Bedrock, Vertex, and Foundry users; opt out with CLAUDE_CODE_USE_POWERSHELL_TOOL=0.
Worktree Isolation
Added worktree.bgIsolation: "none" setting to let background sessions edit the working copy directly without EnterWorktree — useful for repos where worktrees are impractical.
New Flags for claude agents
claude agents now accepts --add-dir, --settings, --mcp-config, --plugin-dir, --permission-mode, --model, --effort, and --dangerously-skip-permissions. These apply to the dashboard and to background sessions dispatched from it.
Background Session Improvements
/bgnow preserves--mcp-config,--settings,--add-dir,--plugin-dir,--strict-mcp-config,--fallback-model, and--allow-dangerously-skip-permissionsacross respawn.- Background sessions preserve model and effort level after waking from idle.
- Fixed false-positive worker-stall detection after host sleep or macOS App Nap.
- Background daemon spawn falls back to the running binary when
~/.local/bin/claudelauncher is missing.
Bug Fixes
- Fixed corrupt
.credentials.jsonwith non-arrayscopeshanging the CLI on startup. - Fixed right-click paste in
claude agentson Windows Terminal and WSL. - Stop hooks that block repeatedly now end the turn with a warning after 8 consecutive blocks (override via
CLAUDE_CODE_STOP_HOOK_BLOCK_CAP). - Fixed Esc/Ctrl+C not cancelling a pending
/loopwakeup. - Fixed
/goalevaluator firing while background shells or delegated subagents are running. - Fixed
NO_COLOR/FORCE_COLORinsettings.jsonenv stripping Claude Code's own UI colors — they now apply to subprocesses only. - Fixed agent view spawning repeated PowerShell processes on Windows when listing sessions.
- Fixed
/bgwithout a prompt sending "continue" — the fork now waits for input. - Fixed
--agent <name>not finding plugin-contributed agents withoutplugin:prefix. - Fixed deleting a session from agent view not removing its transcript file.
- Fixed stale-fragment rendering when scrolling in attached background sessions on Windows Terminal.
- Fixed 5xx error messages pointing at status.claude.com instead of naming the configured gateway.
- Worktree cleanup no longer falls back to
rm -rfwhengit worktree removefails. - Fixed background-job sessions on macOS getting "Operation not permitted" for
~/Documents,~/Desktop,~/Downloadseven with Full Disk Access. - Fixed background sessions capturing IDE file references into warm spare input.
- Fixed
claude agents --allow-dangerously-skip-permissionsdefaulting to bypass mode instead of just making it available in the permission cycle.
Shift+Tab Cycle
Shift+Tab in attached agent sessions now includes auto mode in the cycle.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Launch Engine MCP Server Provides 39-Tool Pipeline for Business Validation
Launch Engine is an MCP server that gives Claude a structured pipeline with 39 interconnected SOP tools organized into 5 layers for taking business ideas from concept to validated revenue. The system includes specialized subagents, prerequisite enforcement, and tools for batch evaluation and rapid testing.

Agentic Context Engine: Automated Agent Improvement Loop with 34.2% Accuracy Gain
An open-source tool automates the entire agent improvement loop from trace analysis to fix implementation, achieving 34.2% accuracy improvement on Tau-2 Bench in one iteration. The system uses Claude Code in a REPL environment to analyze failures and decide between prompt or code fixes.

ClaudeClaw: Free Claude Code Plugin for Persistent AI Agents on Messaging Platforms
ClaudeClaw is a free, open-source MIT-licensed Claude Code plugin that runs Claude as a persistent agent on Slack, WhatsApp, and Telegram. It requires Node.js 20+, Claude Code, and your own Anthropic API key, with OS-level sandbox isolation via Anthropic's sandbox-runtime.

Obsidian Integration for Persistent Memory in OpenClaw and Claude Code
A Reddit user demonstrates how connecting OpenClaw and Claude Code to an Obsidian vault creates persistent long-term memory across sessions. The setup automatically links memories, context, project files, and notes, with all instances able to access shared memory when needed.