Claude Code command and telemetry controls

Claude Code added broader shell-command classification, clearer permission-denial reasons, MCP authentication notices, and controls for assistant-response telemetry logging. Why it matters: Coding agents need explicit boundaries before they run commands or emit operational traces. Permission and telemetry defaults now deserve the same review as model choice.

Try this: Review Claude Code settings before the next agent run: enable broad shell-command classification where available and set telemetry environment variables to match the intended logging policy.