Claude Code v2.1.215: /verify and /code-review No Longer Automatic

Claude Code v2.1.215 changes how the /verify and /code-review skills behave. Previously, the agent ran these automatically during certain operations. Starting with this release, they no longer run on their own — you must explicitly invoke them with the /verify or /code-review commands.
Key Changes
- Automatic execution removed: Claude will no longer trigger
/verifyor/code-reviewunprompted. - Manual invocation required: Use
/verifyor/code-reviewdirectly in the chat when you want those skills to run.
Impact on Workflow
This change gives you tighter control over when verification and code review happen. If you relied on the automatic behavior, you'll need to add explicit commands to your prompts or scripts. For users who found auto-verification disruptive, this is a welcome reduction in noise.
Updating
If you're on an older version, update Claude Code to v2.1.215 to apply this change. You can check your version and update via your package manager or download from the releases page.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Uber Burned Through Its Yearly Claude Code Budget in 4 Months — Here's What That Means
Uber reportedly exhausted its entire Claude Code budget for the year by month four. The post breaks down why usage patterns broke the subscription model and what devs can learn about setting boundaries with AI coding agents.

Google's TimesFM 2.5: 200M-parameter time-series model with 16k context
Google Research released TimesFM 2.5, a 200M-parameter decoder-only foundation model for time-series forecasting with 16k context length and continuous quantile forecasting up to 1k horizon.
Claude Code v2.1.227 Fixes Feature Flag Subscriptions and Bash Errors in CI
Claude Code v2.1.227 fixes feature-flag evaluation with expired tokens, Bash failures in claude-code-action, and improves slash-command menu accessibility.

Mercor Breach: 4TB of Voice Samples + IDs Stolen – What Attackers Can Do Now
4TB of voice recordings paired with government IDs stolen from 40,000 Mercor contractors. Attackers can clone voices from 15 seconds of clean audio and bypass bank voiceprint verification, deepfake calls, and insurance fraud.