0.106.0-cometix

Release 0.106.0-cometix

New Features

• Added a direct install script for macOS and Linux and publish it as a GitHub release asset, using the existing platform payload (including codex and rg) (openai#12740)
• Expanded the app-server v2 thread API with experimental thread-scoped realtime endpoints/notifications and a thread/unsubscribe flow to unload live threads without archiving them (openai#12715, openai#10954)
• Promoted js_repl to /experimental, added startup compatibility checks with user-visible warnings, and lowered the validated minimum Node version to 22.22.0 (openai#12712, openai#12824, openai#12857)
• Enabled request_user_input in Default collaboration mode (not just Plan mode) (openai#12735)
• Made 5.3-codex visible in the CLI model list for API users (openai#12808)
• Improved memory behavior with diff-based forgetting and usage-aware memory selection (openai#12900, openai#12909)

Bug Fixes

• Improved realtime websocket reliability by retrying timeout-related HTTP 400 handshake failures and preferring WebSocket v2 when supported by the selected model (openai#12791, openai#12838)
• Fixed a zsh-fork shell execution path that could drop sandbox wrappers and bypass expected filesystem restrictions (openai#12800)
• Added a shared ~1M-character input size cap in the TUI and app-server to prevent hangs/crashes on oversized pastes, with explicit error responses (openai#12823)
• Improved TUI local file-link rendering to hide absolute paths while preserving visible line/column references (openai#12705, openai#12870)
• Fixed Ctrl-C handling for sub-agents in the TUI (openai#12911)

Documentation

• Fixed a stale sign-in success link in the auth/onboarding flow (openai#12805)
• Clarified the CLI login hint for remote/device-auth login scenarios (openai#12813)

Chores

• Added structured OTEL audit logging for embedded codex-network-proxy policy decisions and blocks (openai#12046)
• Removed the steer feature flag and standardized on the always-on steer path in the TUI composer (openai#12026)
• Reduced sub-agent startup overhead by skipping expensive history metadata scans for subagent spawns (openai#12918)

Changelog

• openai#12046 feat(network-proxy): add embedded OTEL policy audit logging @mcgrew-oai
• openai#12712 Promote js_repl to experimental with Node requirement @fjord-oai
• openai#12715 Add app-server v2 thread realtime API @aibrahim-oai
• openai#12795 Revert "fix(bazel): replace askama templates with include_str! in memories" @jif-oai
• openai#12791 Handle websocket timeout @pakrym-oai
• openai#12800 fix: enforce sandbox envelope for zsh fork execution @bolinfest
• openai#12802 Propagate session ID when compacting @rasmusrygaard
• openai#12732 feat(app-server): add ThreadItem::DynamicToolCall @owenlin0
• openai#12807 Add simple realtime text logs @aibrahim-oai
• openai#12805 Update Codex docs success link @etraut-openai
• openai#12809 fix: harden zsh fork tests and keep subcommand approvals deterministic @bolinfest
• openai#12808 make 5.3-codex visible in cli for api users @sayan-oai
• openai#10954 feat(app-server): thread/unsubscribe API @owenlin0
• openai#12806 only use preambles for realtime @aibrahim-oai
• openai#12830 Revert "only use preambles for realtime" @aibrahim-oai
• openai#12721 Revert "Ensure shell command skills trigger approval (openai#12697)" @celia-oai
• openai#12831 only use preambles for realtime @aibrahim-oai
• openai#12735 Enable request_user_input in Default mode @charley-oai
• openai#12814 feat: scope execve session approvals by approved skill metadata @bolinfest
• openai#12026 Remove steer feature flag @aibrahim-oai
• openai#12740 Add macOS and Linux direct install script @efrazer-oai
• openai#12838 Use websocket v2 as model-preferred websocket protocol @pakrym-oai
• openai#12811 Revert "Add skill approval event/response (openai#12633)" @celia-oai
• openai#12758 feat: include available decisions in command approval requests @bolinfest
• openai#12824 Disable js_repl when Node is incompatible at startup @fjord-oai
• openai#12813 Clarify device auth login hint @xl-openai
• openai#12848 Try fixing windows pipeline @pakrym-oai
• openai#12856 Attempt 2 to fix release @pakrym-oai
• openai#12744 Skip system skills for extra roots @xl-openai
• openai#12857 Reduce js_repl Node version requirement to 22.22.0 @fjord-oai
• openai#12865 Fix release build take @pakrym-oai
• openai#12705 Hide local file link destinations in TUI markdown @pash-openai
• openai#12823 Enforce user input length cap @etraut-openai
• openai#12417 core: bundle settings diff updates into one dev/user envelope @charley-oai
• openai#12884 chore: new agents name @jif-oai
• openai#12885 nit: captial @jif-oai
• openai#12870 tui: restore visible line numbers for hidden file links @pash-openai
• openai#12684 Add rollout path to memory files and search for them during read @wendyjiao-openai
• openai#12901 chore: better awaiter description @jif-oai
• openai#12900 feat: memories forgetting @jif-oai
• openai#12905 chore: clean DB runtime @jif-oai
• openai#12918 Skip history metadata scan for subagents @daveaitel-openai
• openai#12871 split-debuginfo @pakrym-oai
• openai#12909 feat: use memory usage for selection @jif-oai
• openai#12887 fix: do not apply turn cwd to metadata @jif-oai
• openai#12911 fix: ctrl c sub agent @jif-oai
• openai#12873 Use model catalog default for reasoning summary fallback @pakrym-oai

Cometix

• Carried forward: bundled models.json visibility preserved during remote model merge.
• Carried forward: thin LTO + opt-level=s for cometix releases.
• Carried forward: context segment placeholder, dead_code suppressions, snapshot updates.
• Removed upstream install.sh staging step from release workflow (hardcoded openai/codex URLs).