2.1.207
Highlights:
- Auto mode enabled by default on Bedrock/Vertex AI/Foundry; disable with disableAutoMode to stop automation
- Fix terminal freeze and key lag when streaming long lists/tables/paragraphs/code, restoring responsiveness
- Noninteractive runs stop silently recording consent for remote-managed settings if consent dialog wasn't shown
CLI changelog
- Fixed the terminal freezing and keystrokes lagging while streaming responses containing very long lists, tables, paragraphs, or code blocks
- Fixed remote managed settings from a non-interactive run (claude -p, the SDK) being permanently recorded as consented without ever showing the security consent dialog
- Fixed spurious prompt-injection warnings triggered by benign system-generated conversation updates
- Fixed the auto-updater overwriting a custom launcher script or symlink at ~/.local/bin/claude on every release; /doctor now reports an externally managed launcher
- Fixed compound commands with cd prompting for permission when the only output redirect was to /dev/null
- Fixed the transcript jumping above the start of the answer when a response finishes streaming
- Fixed extensions.worktreeConfig being left in the repo's .git/config (breaking go-git tools like tea) after the last worktree.sparsePaths worktree was removed
- Fixed malformed bracket patterns in rules globs, skill paths, .ignore, and .worktreeinclude breaking file reads, file suggestions, and worktree creation
- Fixed a crash loop in agent teams where a malformed teammate mailbox message caused repeated errors every second until the mailbox file was manually deleted
- Fixed background sessions auto-named by accepting a plan not showing that name on their agent-view row
- Fixed background sessions that entered a git worktree resuming blank after a cold reopen from the agent list
- Fixed Remote Control task status updates being lost when the connection recovered from a network interruption or credential refresh
- … +12 more CLI changelog entries
Source: https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md#21207
Metadata
-
Time since 2.1.206 release: 1d 19h 54m 31s
-
Bundle file size: +120.5 kB (+0.3%)
-
Prompt files: +3 (+5.9%)
-
Prompt tokens: +11,939 (+9.0%)
Prompt token mix by kind:
- system-reminder: 57.5%→54.7%
- tools: 41.8%→41.7%
- system: 0.4%→3.3%
- agent: 0.2%→0.2%
Files:
- metadata: 288552e/meta/metadata.md#diff-a08dfa163baef97ea4f56d29009ac65f5f7dbc46e3a58e80d7aa06a2a2520899L10-R10
- prompt stats: 288552e/meta/prompt-stats.md#diff-fa8e31336ece71d159f37b6fd4ed0c61939849140bd5c08c7ca793b5b2ae7f58L1-R1
System prompt changes
Notable changes:
- Claude no longer gets told that tags in messages/tool results are injected by the harness (not user text). This weakens guidance on how to interpret those tags, increasing the chance Claude treats them like user-provided content instead of system-side annotations.
- Claude now sees an extra Bun hot-update warning about import.meta.hot.accept placed ahead of the main security/harness rules. This adds non-instructional, task-specific noise that can divert attention or trigger irrelevant troubleshooting guidance in otherwise unrelated conversations.
Links:
- e54e1e4/cc-prompt.md#diff-b0a16d13c25d701124251a8943c92de0ff67deacae73de1e83107722f5e5d7f1L63-R64
- e54e1e4/cc-prompt.md#diff-b0a16d13c25d701124251a8943c92de0ff67deacae73de1e83107722f5e5d7f1R59
Other prompt changes
- Adds a system reminder directing the assistant to help with software engineering tasks, use persistent file-based memory, prefer file-based tools over shell, and use Agent/subagents for broad code exploration.
Links: