Claude Code v2.1.144: Background Sessions, /model Scoping, and 15s Startup Timeout

Claude Code v2.1.144 is out, bringing several quality-of-life improvements and critical bug fixes. The release focuses on better background session management, terminal rendering fixes, and more robust error handling.
Background Sessions Get First-Class Support
- /resume background sessions — sessions started via
claude --bgor agent view now appear alongside interactive ones, marked with abglabel. You can list them withclaude sessionsor pick from the session list. - Background subagent completion notifications now show elapsed duration (e.g.,
Agent completed · 3h 2m 5s). /bgand← -detachnow preserve the working directory.
/model Scoped to Current Session
/model now changes the model for the current session only. To set a default for new sessions, press d in the model picker. Resumed sessions keep the model they were using, rather than picking up another session's /model choice.
Startup Timeout Reduced
Fixed a bug where startup could hang up to 75 seconds when api.anthropic.com is unreachable (captive portal, firewall, VPN issues). Side-channel API calls now time out after 15 seconds.
Terminal Rendering Fixes
- Garbled output after missed window-resize events (e.g., dragging a VS Code split-pane divider) now self-heals on the next frame instead of requiring
Ctrl+L. - Progressive terminal display corruption (stale/garbled glyphs) in very long sessions — now cleared automatically instead of requiring terminal resize or restart.
- Reduced spinner animation color count to minimize rendering glitches in VS Code.
- Windows: scrolling in attached background sessions fixed (PgUp/PgDn, mouse wheel,
Ctrl+Otranscript navigation). - Windows: fix for ghost characters at left edge when switching panes in Agent View with CJK content.
Other Notable Fixes
- File extension mismatch: Reading a file whose image extension doesn't match its contents (e.g., HTML saved as
.png) now falls back to text instead of crashing the conversation. - Search tool errors reduced:
head/tailfile views now satisfy the read-before-edit check;egrep,fgrep,git grep,git diffwith exit code 1 (no matches) no longer reported as command failures. - MCP pagination: Fixed paginated
tools/listresponses only returning first page, silently dropping tools. - MCP unsupported MIME: Images with unsupported MIME types (e.g., SVG) now saved to disk and referenced in tool result instead of breaking the conversation.
- Skill directory file descriptors: Non-
.mdfiles in skill directories no longer trigger skill reloads, preventing file descriptor exhaustion. - macOS Full Disk Access: Background sessions crashing with
exit 1 before initwhen project is under a Full Disk Access–protected folder (regression in v2.1.143) fixed. - Plugin not cached: Plugins enabled in own settings no longer show
not cachederrors on first load; plugins from project's.claude/settings.jsonnow show actionable install hint. - MCP config errors:
claude mcp listnow shows configuration errors when.mcp.jsoncan't be parsed (e.g., using VS Code'sserverskey instead ofmcpServers).
Command Line Changes
/extra-usagerenamed to/usage-credits; old name still works./pluginbrowse and discover panes now show when a plugin was last updated.
This release is available now for all Claude Code users. If you use MCP servers heavily or run long background sessions, the rendering and pagination fixes alone are worth the upgrade.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Top 6 Open Source Claude Skills (April 15 – May 3)
Six open-source Claude skills from the last 15 days: brand-alchemy, npm-downloads-to-leads, hyperframes, email-newsletter, pricing, and more. Detailed breakdown of each skill's functionality.

Open-source AI job search system built with Claude Code evaluates offers, generates tailored resumes
A developer open-sourced a Claude Code project that turns your terminal into a job search command center. The system evaluates job offers across 10 dimensions, generates ATS-optimized PDF resumes, scans 45+ company career pages, and includes 14 skill modes.

Agent-Xray: Open-source tool for debugging AI agent failures from trace logs
Agent-Xray is an MIT-licensed open-source tool that analyzes AI agent trace logs to classify failures into categories like spin, tool_bug, and early_abort, and includes an enforcement mode to test fixes against adversarial challenges.

Distillery: A Claude Code Plugin for Persistent Team Context
Distillery is a plugin for Claude Code that provides teams with shared, persistent context across sessions and people. Version 0.2.0 adds hybrid search, auth audit logging, and uv support.