Claude Code v2.1.199 Patches 20+ Bugs: SSL, Subagents, Daemon Crashes Fixed

Claude Code v2.1.199 is out with over 20 bug fixes addressing streaming reliability, subagent error handling, SSL certificate issues, and daemon stability. This release is a serious patch for anyone running Claude Code in production or multi-agent workflows. Here's what changed.

Key Fixes

• SSL certificate errors: Previously, TLS-inspecting proxies or missing NODE_EXTRA_CA_CERTS would cause repeated retries before showing guidance. Now the error fails immediately with the fix hint.
• Streaming partial output retained: If the API emits a mid-stream overloaded/server error after partial output, the partial is kept with an incomplete-response notice instead of being discarded.
• Subagent silent failures fixed: Subagents cut off by rate limits or server errors now return partial work to the parent agent instead of failing silently. API errors (e.g., usage limit reached) are reported to the parent agent, not passed as successful results.
• Linux daemon crash loop: The background-agent daemon on Linux was killing itself and all running agents every ~50 seconds after an unclean shutdown left a corrupted worker record. This is now fixed.
• macOS SSH cold-start: Background agents failing to cold-start over SSH with 'Could not switch to audit session' (regression in 2.1.196) now work correctly.
• claude stop being silently undone when it raced a background-agent respawn — the respawn now honors the stop.
• Background job progress indicators no longer stall for minutes while the job runs long commands.
• Memory-starved machines now show a specific low-memory error with suggestions to free resources, instead of a generic error.
• SendMessage misrouting: when a re-spawned agent reuses a previous agent's name, the tool now detects the mismatch and asks the caller to retarget.
• Plan mode no longer prompts for read-only browser_batch calls — they are auto-allowed. State-changing browser tool calls now correctly prompt.

Other Fixes

• Stacked slash-skill invocations (/skill-a /skill-b do XYZ) now load all leading skills (up to 5), not just the first.
• Transient server rate-limit errors (429s unrelated to your usage limit) are now retried automatically with backoff for subscribers instead of failing the turn.
• CLAUDE_CODE_RETRY_WATCHDOG raises default retry count for non-capacity transient errors to 300 and lifts the cap of 15 on CLAUDE_CODE_MAX_RETRIES.
• Typing /model or /fast while viewing a subagent no longer silently opens the lead's model picker — a notice explains the command applies to the lead.
• SessionStart, Setup, and SubagentStart hooks now show stderr on exit code 2 in the transcript.
• claude --dangerously-skip-permissions daemon <subcommand> correctly runs the subcommand instead of being treated as a chat prompt.
• Opening or resuming a session with no new messages no longer needlessly grows the transcript file.
• Backgrounding a session with ← or /background does not drop its /color from the agent view row.
• Resetting a corrupted config file from the startup recovery dialog now backs up the file first before destroying it.
• Claude in Chrome no longer repeatedly opens the reconnect page when sessions run from different builds or config directories.
• Remote sessions no longer flap between Working and Idle in the agent view when a background agent completes.
• Idle subagents now collapse into an expandable summary row instead of vanishing while other subagents work.
• claude agents session rows show PR links as bare #N without the 'PR' label.

Who It's For

Any developer running Claude Code in multi-agent setups, behind corporate proxies, on Linux servers, or relying on background agents will want to upgrade immediately.