Claude Code 2.1.330 adds 'measure twice, cut once' confirmation rule for destructive actions
ClaudeCodeLog · x · 2026-11-21
Claude Code 2.1.330 ships a notable system prompt change: a "measure twice, cut once" rule.
- Claude must now weigh reversibility and blast radius, and ask for confirmation before destructive, hard-to-reverse, shared-state, or public actions — including rm -rf, reset --hard, force-push, PR comments, Slack/email sends, and third-party uploads — with git status/stash checks required.
- The CLI surface also saw many changes: new model identifiers like claude-code-android-emulator and claude-code-ios-simulator, plus removal of legacy env vars such as ANTHROPICDEFAULTHAIKUMODEL and AWSBEARERTOKENBEDROCK.
Related event: Claude Code 2.1.330 Adds Dangerous-Operation Confirmation Rule(4 posts)→
More from coding & agent
- Coinbase CEO details agentic workflow: 7-8 lead agents, 1 chief of staff, sub-agent squads — davidsenra · 2026-11-21
- AWS Marketplace rolls out AI agents that can autonomously buy software for companies — jeremyakahn · 2026-11-21
- CT-Bench, a benchmark for LLM agent cooperation via self-negotiated contracts, accepted at COLM 2026 — maxhkw · 2026-11-21
- Google's Pro² Agent hits 76% medal rate on MLE-Bench, a new record — dair_ai · 2026-11-21
- Cole Medin Demos His AI Software Factory Shipping Ideas to Working Code Live — Cole Medin · 2026-11-21
- Open-source PlaceBench tests 6 voice assistants on 150 real-place questions: Dots 92, Ask Maps 49 — whowasntright · 2026-11-21