v0.4.3

What's Changed

• Update demo.gif
• providers: refresh model catalogue and extend fetch script
• script: open changelog in editor and add model refresh to release
• config: default general agent to claude-opus-4-6
• docs: document todo list as user-facing in AGENTS.md and tool schema
• llm: fix OpenAI Codex backend error detail and store parameter
• llm: disable adaptive thinking for Claude Haiku models
• Merge branch 'main' of github.com:get-vix/vix
• config: add credential store with per-method provider auth
• ui: add sessions-list loading spinner for busy sessions
• telemetry: rotate session id and track turns and session end
• ui: group slash menu and advertise skills before brain init
• daemon: shorten daemon exit grace period to 3s
• daemon: allow mach-register and broaden exec in Seatbelt profile
• docs: document vixd detection caveat under the sandbox
• build: add Docker test harness for clean-container runs
• Merge pull request #24 from pankaj4u4m/bedrock-provider
• bedrock: wire base_url from providers registry in endpoint construction
• gofmt: format all source files
• ui: unify F-key tab switching and clear unread latch on read
• update: add in-app self-update with daily release check
• Merge upstream get-vix/vix main into bedrock-provider branch
• remove ratelimit
• fix: log rate limiter activation at vixd startup
• fix: address code review — harden Bedrock adapter and standardise rate limiting
• fix: include bedrock in pre-flight credential check
• feat: add --model flag to override agent frontmatter per session
• fix: make Bedrock RPM rate limit opt-in via VIX_BEDROCK_RPM env var
• feat: add Amazon Bedrock provider with ephemeral prompt caching