[Release] anthropics/claude-code v2.1.139: v2.1.139

Claude Code v2.1.139 introduces significant new features including an agent view for managing multiple sessions, a `/goal` command for multi-turn task completion, and enhanced MCP server support with environment variables. The release includes numerous UX improvements like transcript navigation shortcuts, plugin token cost estimation, and fixes for critical issues such as credential deadlocks, memory leaks in MCP servers, and terminal interaction bugs.

Key Points

• •Agent view (Research Preview) provides a unified list of all Claude Code sessions with status tracking — access via `claude agents` command
• •/goal command enables setting completion conditions that persist across multiple turns with live metrics (elapsed time, turn count, token usage)
• •MCP stdio servers now receive `CLAUDE_PROJECT_DIR` environment variable, enabling plugin configs to reference project paths via `${CLAUDE_PROJECT_DIR}`
• •Plugin token cost estimation available via `claude plugin details <name>` showing component inventory and per-session token projections
• •Transcript view navigation enhanced with keyboard shortcuts (`?` for help, `{`/`}` for prompt jumping, `v` for shortcut panel toggle)
• •Hook `args` field now supports exec form (direct command spawning without shell) eliminating path quoting requirements
• •PostToolUse hook `continueOnBlock` option allows feeding rejection reasons back to Claude for continued processing
• •Fixed critical deadlock preventing `claude auth` commands when credentials expire and `forceRemoteSettingsRefresh` policy is active
• •MCP server response bodies capped at 16 MB per SSE frame to prevent unbounded memory growth from non-protocol data streams
• •Subagent API requests now include `x-claude-code-agent-id` and `x-claude-code-parent-agent-id` headers for improved tracing and observability

Workflow Diagram

Concepts