Codex CLI v0.170.0 is a OpenAI release published 2026-11-18 with 58 listed changes. Summary and full notes below, extracted from the official notes.
One free email: every breaking change, retirement and price change across 23 coding agents. No spam, unsubscribe anytime.
Subscribe free← All Codex CLI releases and breaking changes
## New Features - Added WebRTC negotiation for realtime voice calls in the TUI. (#61172) - Added automatically refreshed authentication headers for OpenTelemetry HTTP exporters. (#61028) - Added experimental support for using identically named MCP servers across environments. (#60905) ## Bug Fixes - Oversized WebSocket requests now fall back to HTTP without unnecessary retries. (#60840, #60841) - Fixed MCP OAuth credential races during cancellation, login, and logout. (#60849, #60981) - Improved automatic reconnection for Cloud conversations and preserved active recovery turns when clients reconnect. (#60969, #61014) - Fixed Windows file replacement, sandbox permission preservation, and runtime registration recovery. (#60872, #60943, #60944) - Fixed permission updates and child threads losing existing sandbox restrictions. (#60936, #61134) - Removed duplicate file search results and fixed thread search snippets for queries containing whitespace. (#60845, #60846) ## Documentation - Updated sandbox documentation to explain permission profiles and filesystem rules. (#61133) ## Changelog Full Changelog: https://github.com/openai/codex/compare/rust-v0.169.0...rust-v0.170.0 - #60770 Derive history and notes tool schemas from typed requests @euroelessar - #60771 Preserve model-facing ID allocations in saved history @euroelessar - #60773 Resolve saved Daybreak choices for each subagent turn @sayan-oai - #60784 Centralize synchronous Guardian reviewer selection @felixxia-oai - #60839 Preserve retained assistant revisions across rollout replay @euroelessar - #60840 Fall back to HTTP immediately for oversized WebSocket requests @charliemarsh-oai - #60841 Honor advertised WebSocket message size limits @charliemarsh-oai - #60842 Make TUI thread titles resilient to generation failures @etraut-openai - #60845 Deduplicate file search results from overlapping roots @etraut-openai - #60846 Fix thread search snippets for whitespace-containing queries @etraut-openai - #60848 Enable incremental tool declarations by default for GCP @akira-oai - #60849 Protect MCP OAuth persistence from stale credential writes @mzeng-openai - #60871 Wait for the in-process initialization handshake to finish @goodman-oai - #60872 Fix Windows atomic writes and refresh build maintenance @zm-oai - #60882 Stabilize tracing in the cancellation timing test @bc-openai - #60897 Use the Responses API for issue label suggestions @GrantBirki - #60898 Include bundled bubblewrap in the root metadata test fixture @sayan-oai - #60904 Version answer display instructions and restrict unsupported payloads @psachdeva-oai - #60905 Keep same-named MCP servers separate across environments @sayan-oai - #60914 Reserve Guardian context budget for selected Code Mode exec results @felixxia-oai - #60929 Preserve Command Center task ownership across backend switches @bc-openai - #60936 Preserve legacy sandbox templates in TUI permission conversions @ioanbudea-OAI - #60942 Preserve thread-local model ID state across replay and compaction @euroelessar - #60943 Preserve file permissions during repeated Windows sandbox setup @zm-oai - #60944 Reconcile Windows runtime registration after service privilege errors @zm-oai - #60948 Report biometric type in user verification responses @bdavis-oai - #60962 Refactor shortcut overlay binding construction @bc-openai - #60964 Validate the issue labeler's API key before sending requests @GrantBirki - #60969 Reconnect idle Cloud conversations after transient stream loss @bc-openai - #60972 Lowercase service tier names in the model status line @bc-openai - #60977 Preserve allocated response item IDs in requests @euroelessar - #60981 Make reactive MCP OAuth refreshes cancellation-safe @mzeng-openai - #61014 Preserve active recovery turns when clients reconnect @apanasenko-oai - #61017 Record reasoning replay removal metrics @akira-oai - #61018 Extract HTTP headers helper execution into a shared crate @abhinav-oai - #61028 Add refreshable headers for OTLP HTTP exporters @abhinav-oai - #61129 Simplify non-curated plugin cache refreshes @smurching-oai - #61133 Use permission profiles in sandbox tests and documentation @ioanbudea-OAI - #61134 Normalize permission updates at session settings admission @ioanbudea-OAI - #61136 Compare resume sandbox overrides using permission profile labels @ioanbudea-OAI - #61138 Centralize filesystem helper sandbox preparation @viyatb-oai - #61141 Preserve multiline status lines in unfocused chat panes @fcoury-oai - #61144 Remove the legacy sandbox policy setter from `Config` @ioanbudea-OAI - #61153 Fix code-mode session admission retries for proxy HTTP 503 responses @seanh-oai - #61156 Ignore Hickory DNSSEC advisories while DNSSEC is disabled @swastik-oai - #61163 Reuse Guardian reviewer history for direct reviews @swastik-oai - #61167 Refactor TUI sidebar layout and focus handling around panels @fcoury-oai - #61172 Negotiate realtime voice calls in the TUI @zahan-oai