CLAUDE CODE 2.1.111 LANDS OPUS 4.7 XHIGH, AUTO MODE UPGRADES, AND CLOUD ULTRAREVIEW; 2.1.112 HOTFIX FOLLOWS
Anthropic shipped a sizable Claude Code update with smarter model controls, fewer permission nags, and a new multi-agent cloud code review. The 2.1.111 release...
Anthropic shipped a sizable Claude Code update with smarter model controls, fewer permission nags, and a new multi-agent cloud code review.
The 2.1.111 release adds Opus 4.7 “xhigh” effort, a live /effort slider, Auto mode for Max subscribers on Opus 4.7, and a new /ultrareview that runs parallel, cloud-based multi-agent PR reviews release notes. It also cuts friction: fewer read-only Bash prompts, a PowerShell tool rollout on Windows, and better /setup for Vertex/Bedrock including a “with 1M context” option. A fast follow, 2.1.112, fixes an auto-mode availability error hotfix.
Two ecosystem gotchas surfaced the same day. A community fix in claude-mem restores broken observations for Claude Code 2.1.109+ due to an Agent SDK flag quirk v12.1.6. And a VS Code startup timeout blamed on MCPs was actually a blocking SessionStart hook; removing the long-running hook resolves it debug write-up.
Cloud multi-agent PR reviews (/ultrareview) and better model tuning can tighten review loops and raise code quality.
Permission-prompt trimming and PowerShell support reduce day-to-day friction for mixed OS teams.
-
terminal
Run /ultrareview on a few real PRs and compare issue catch rate and review latency versus your current code-review process.
-
terminal
Benchmark Opus 4.7 effort levels (high, xhigh, max) on a representative repo task for latency, cost, and output quality.
Legacy codebase integration strategies...
- 01.
If you use claude-mem, upgrade to v12.1.6 to restore observations on Claude Code 2.1.109+ and avoid silent data loss.
- 02.
Audit ~/.claude/settings.json SessionStart hooks in VS Code; remove long-running commands that can block initialization.
Fresh architecture paradigms...
- 01.
Adopt /ultrareview in your PR template for an automated first pass, then gate human review on AI findings.
- 02.
Consider Vertex/Bedrock setup with the 1M context option for large repo reasoning or wide test-log ingestion.
Get daily CLAUDE-CODE + SDLC updates.
- Practical tactics you can ship tomorrow
- Tooling, workflows, and architecture notes
- One short email each weekday