Claude Code 2.1.176 fixes language, model, and remote issues

Claude Code 2.1.176 fixes language, model, and remote issues

Claude Code 2.1.176 now creates session titles in the same language as the conversation. A language setting can lock titles to one chosen language. A new footerLinksRegexes setting lets teams show link badges in the footer when links match a set pattern. Bedrock credentials are now cached until their real expiration time instead of always using a fixed one-hour window. Model allowlist rules are stricter: alias-based model choices can no longer be redirected to a blocked model through ANTHROPIC_DEFAULT_*_MODEL environment variables, and /fast will refuse to switch if it would move outside the allowed list. Auto mode no longer fails on Fable 5 for organizations that do not have Opus 4.8 enabled, because it falls back to the best Opus model available. Fixes also cover hook path matching for file tools, Linux sandbox startup with certain settings links, copy behavior inside tmux over SSH, and Remote Control cases where web or mobile connections silently changed the session model.

Key points

  • Session titles now follow the conversation language, with a setting to pin one language.
  • footerLinksRegexes adds pattern-based link badges in the footer.
  • Bedrock credentials are cached until their actual expiration time.
  • Model allowlist enforcement was tightened for aliases, environment variables, and /fast.
  • Fixes also cover hook path matching, Linux sandbox startup, tmux copy behavior, and Remote Control model switching.

Quick term guide

footerLinksRegexes
A Claude Code setting for showing footer link badges when links match chosen text patterns.
credentials
Secret keys or tokens used to access an account or service.
credential
Information that proves permission to use an account, card, or service.
environment variables
Settings passed to a program when it runs.
environment variable
A setting stored outside your code that a program reads at startup, commonly used for API keys and configuration options.
Auto mode
A setting where the AI keeps working through multiple steps on its own, without stopping to ask for your approval
reliability
How consistently a tool works without failing or behaving unexpectedly.
managed setting
A setting controlled by an admin rather than each user.
Read original