## Why
Optimized `Array.prototype.sort` can write objects into incompatible array element storage when receiver element kinds differ and the comparator mutates the array.
## What changed
- Upgrade `v8` to `152.2.0`, the Bazel V8 source to `15.2.124.1`, and ICU data and initialization to version 78.
- Backport element-kind agreement checks for sort inlining in TurboFan and Maglev. Apply the patch to Bazel builds and Windows MSVC source builds, and include it in source-build cache keys and canary triggers.
- Refresh pinned sandbox archives and bindings, extend Bazel prebuilt coverage to musl Linux, and use matching sandbox bindings on Windows MSVC.
- Update Bazel integration patches and runtime dependencies for the new V8 release, and download the pinned Windows build tools in source-build workflows.
## Testing
Include the upstream sort regression test covering mixed element kinds, comparator mutation, and continued optimization for matching kinds. Add a test requiring canary and Windows source builds when the backport changes, and update artifact-fetch tests for the new version.
GitOrigin-RevId: 138b7821cddd25466c10aced1dd22be24837caf0
## Why
A catchable exception lets JavaScript continue after `exit()`, including in `catch`, `finally`, and queued promise callbacks. Exiting should finish the cell successfully without further output or state changes.
## What changed
- Terminate V8 execution when `exit()` is called and complete the cell even if its module promise remains pending.
- Guard host callbacks against effects after exit, including when getters or value conversion request exit.
- Preserve stored writes made before exit and suppress errors when exit occurs during exception formatting.
## Testing
Add regression tests for synchronous and asynchronous exit, timers, parallel promises, value conversion, exception getters, and resuming after a yield and tool response. Verify that only prior output and stored writes survive, and separately exercise host guards while JavaScript continues running.
GitOrigin-RevId: e9487ba65ef0358612c5009bd26953b786a12798
## Why
Reentering V8 during termination can replace uncatchable cancellation with a catchable JavaScript exception, allowing a script to continue after cancellation.
## What changed
Skip JSON conversion and exception throwing while V8 execution is terminating, preserving cancellation through `image()` getters and JSON fallback.
## Testing
Add regression coverage for cancellation during JSON fallback and reentrant image getters, including prompt termination through the standalone host and continued usability of another session. Extend serialization error tests to cover failing image getters alongside circular JSON.
GitOrigin-RevId: f03dd131a62724106fad96111292ddf5b9dc99f6
## Why
JavaScript objects containing Serde's reserved JSON keys can be reinterpreted during deserialization, allowing nested `RawValue` markers to bypass the parser's normal recursion limit.
## What changed
Reject `$serde_json::private::RawValue` and `$serde_json::private::Number` object keys before passing V8's serialized JSON to `serde_json`. Report a `reserved JSON object key` serialization error while continuing to allow marker text in strings and longer keys.
## Testing
Add runtime regression coverage for nested and escaped marker keys, recursion limits, normal JSON conversion, and session recovery without overwriting stored values. Extend the host-backed integration test to verify that stored values remain available across turns after rejection.
GitOrigin-RevId: 27ddce08e1ffe98ec9dffe02473a815ceae2cc89
## What changed
Consume enabled tool definitions and borrow their metadata when installing `tools` and `ALL_TOOLS`, avoiding copies of the tool catalog. Retain only tool names and kinds in runtime callback state, preserving each cell's original callback indices for its lifetime.
GitOrigin-RevId: 62eaf62abce6de1a0cd5cdb3332fbed754a23993
## What changed
Expose two JavaScript globals for consuming results as inputs settle:
- `as_settled(promises)` returns an async iterator of fulfilled and rejected records in settlement order, preserving queued results while the consumer is busy.
- `stream_settled(promises, emit)` delivers those records to a callback, awaiting each invocation and propagating callback errors.
Records include an `index`, a `status`, and either a `value` or a `reason`. Iterable inputs use positional indices; `Map` inputs retain their keys. Both helpers accept promises, thenables, and plain values.
## Testing
Add runtime coverage for ordering, input types, object identity, callback behavior, early iteration exit, and termination. Add stdio and gRPC host tests for partial results across waits, plus a scenario snapshot verifying fulfilled and rejected records reach the model.
GitOrigin-RevId: cddc9f3679ae04f2421e154caa0417afb15f550d
## Why
Tokio's paused clock can auto-advance while tests await the V8 thread. The service tests need explicit control over time when checking yield limits and grace periods.
## What changed
- Hold virtual time with a blocking-task guard that is released when dropped.
- Replace bounded polling loops with tool-start notifications and direct awaits of response tasks.
- Return an error from the test helper if the cell closes before a nested tool starts.
## Testing
Add tests covering tool-start acknowledgement when a cell throws or is terminated before invoking a nested tool, including assertions on the resulting cell responses.
GitOrigin-RevId: c221c0501ec0354052102c0211cc651008babff3
## Why
Yield signals and queued runtime events can detach an observer during termination, before the remaining output has been drained.
## What changed
Track the active yield signal separately from the observer so termination can disable yielding while keeping the observer attached. Ignore `Started`, `Pending`, and `YieldRequested` events during termination, allowing queued output to reach the observer in `CellEvent::Terminated`.
## Testing
Extend the termination regression test with an immediate yield timeout, a canceled yield signal, and queued start, yield, output, and completion events. Assert that both the termination caller and the initial observer receive the queued output in the terminated event.
GitOrigin-RevId: 1c08bdec06838123aaa102268da2f50b63ddea3f
## What changed
Add an optional `preempt` signal to `CodeModeSession::execute` and `CodeModeSession::wait` so callers can end an observation early while the cell continues running and remains available for later waits.
Support the signal in the in-process runtime and across both remote transports: `operation/yield` with the negotiated `yield-observation` capability for stdio, and `YieldObservation` for gRPC. Preserve signals received before gRPC observation registration, and retain normal timeout behavior for older hosts that lack support.
## Testing
Add runtime and transport tests covering early execute and wait yields followed by successful cell completion, yields before observation registration, and suppression of yield frames for older stdio hosts.
GitOrigin-RevId: 3f8049d8f0582be6b8f0832cc03f2d974a0b8f18
## Why
Fixed schema budgets can strip parameter descriptions from large MCP tools or render their Code Mode input types as `unknown`. Allow larger budgets to preserve that guidance.
## What changed
- Add `mcp_servers.<name>.tool_input_schema_max_bytes` to configure the UTF-8 byte threshold for best-effort input schema compaction, retaining the 5,000-byte default.
- Add `features.code_mode.tool_input_schema_max_bytes` for rendered input types. Use the largest of the 16,000-byte default, the Code Mode setting, and the ordinary MCP server's explicit setting.
- Apply rendering budgets to prompt declarations and the runtime tool catalog, and avoid reusing cached declarations after the Code Mode budget changes.
## Testing
Add coverage for UTF-8 compaction thresholds, budget precedence, configuration round trips, and preservation of parameter guidance in both the Code Mode prompt and `ALL_TOOLS`.
GitOrigin-RevId: 0ef2823a614d4794e636f518b09c0b03a5239b45
## Why
Creating a cell deep-copies every stored JSON value, including values the cell never loads. Storing a value also duplicates its payload between the cell's local state and pending writes.
## What changed
Use `Arc<JsonValue>` throughout the session store, cell runtime, and completion writes to share immutable payloads. Keep a separate map snapshot for each cell so later commits cannot change its view of stored values.
GitOrigin-RevId: 4535f0f2ed0c8f8cbfb8d24cf1cb44cbc74472b5
## Why
The pinned V8 can inline `Array.prototype.sort` with incompatible element kinds when a comparator mutates the array, allowing an object to be stored in an integer-elements array.
## What changed
Disable Maglev, Turbolev, and TurboFan array builtin inlining during code-mode runtime initialization until the V8 artifacts include the upstream fix.
## Testing
Add an integration regression test that requests top-tier and Maglev optimization, checks element kinds after comparator mutation, and verifies ordinary numeric sorting still returns `[1,2,3]`.
GitOrigin-RevId: cea38e920245d6a133a5263118dc664fb3e61838
## Why
Each `setTimeout` spawned a sleeping thread that remained alive until its delay elapsed, even after `clearTimeout` or cell completion.
## What changed
Replace per-timer threads with Tokio sleep tasks held by `AbortOnDropHandle`, so removing a timeout or dropping the isolate cancels its task. Enter the caller's Tokio runtime on the code mode runtime thread to support scheduling these timers.
## Testing
Add a regression test using virtual time to verify that cleared timers release their tasks, an awaited timer completes, and cell completion cancels remaining timers without emitting their output.
GitOrigin-RevId: 59c7e57ed226cb32633e2d4c25b778cfab2af423
## Why
A delegate bound at session creation cannot provide different callbacks for cells sharing that session. Each execution needs to retain its own delegate across yields and release it when the cell is cleaned up.
## What changed
- Pass `CodeModeSessionDelegate` to `execute` instead of session creation.
- Route tool calls, notifications, and cell closure callbacks through the execution's delegate in the in-process runtime and the gRPC and stdio transports.
- Retain delegates with pending executions and live cells, releasing them through closure and cancellation cleanup.
## Testing
Add coverage for distinct delegates across yielded cells, gRPC callbacks before cell admission, and delegate release after completion or abandoned execution cleanup. Update transport tests to verify callbacks reach the owning cell's delegate.
GitOrigin-RevId: 7469f52104b993e790a40c65fb800ad02b7fd606
## Why
Tool response delivery handles kept discarded results alive until the cell ended.
## What changed
Add a local V8 handle scope in `resolve_tool_response` so delivery handles are released when response delivery finishes. Live promises continue to retain their results.
## Testing
Add a regression test that verifies a live promise preserves its response through garbage collection and that dropping the promise makes the response collectible before the cell ends.
GitOrigin-RevId: a74e78acae3ec9a465694d7503150aee123792ec
## Why
Passing `undefined` to a tool could fail JSON parsing because V8 stringifies it as the non-JSON text `undefined`.
## What changed
Return no JSON value for JavaScript `undefined`, so explicit `undefined` tool arguments behave like omitted arguments. Attempts to `store` an undefined value report the existing serializability error and preserve the previous stored value.
## Testing
Add regression coverage for storing `undefined` over an existing `null` value, and extend the current-time tool test to cover `{}`, omitted arguments, and explicit `undefined`.
GitOrigin-RevId: 071050ebb2ecc44c8f08453cf8523b4d21728c7e
## Why
Code-mode callbacks can outlive their initial request and run in separate tasks,
so their trace context must be preserved explicitly.
## What changed
- Preserve the execution context for code-mode callbacks and add spans for
nested tool invocations in the runtime and dispatch broker.
- Propagate each tool invocation span as the parent of its streamed gRPC
callback, falling back to the outer execution context when needed.
- Add a `codex.exec_server.process` span for each process lifetime, including
its process ID and completion result, without retaining the request span.
GitOrigin-RevId: a6059e34d895416f5517e51dad6ca0078355adca
## Why
Audio models cannot reliably encode tool-generated clips shorter than 25 ms.
## What changed
- Measure the available audio frames in base64-encoded PCM and IEEE float WAV outputs.
- Replace clips under 25 ms with explanatory text while preserving surrounding Code Mode output.
- Leave clips at or above the threshold and unrecognized audio formats unchanged.
## Testing
- Cover the duration boundary across multiple sample rates and all supported `audio()` input forms.
- Verify bounded and truncated WAV chunks, output after `yield_control()`, and end-to-end mixed text and audio output.
GitOrigin-RevId: 342b4bdbc65db4238580f9cdc372763c2576eca5
## Why
Code mode wall time should measure the host operation itself, without including
client-side response delays or idle time between requests.
## What changed
- Measure each execute, wait, and terminate request in the code mode host.
- Carry the duration through the stdio and gRPC protocols and use it for
model-visible wall time.
- Emit a structured `codex.code_mode.host_timing` event correlated with the
conversation, turn, tool call, and cell.
## Testing
- Cover successful and failed execution timing, delayed response reads,
repeated waits, termination, and missing cells across stdio and gRPC.
- Verify timing survives protocol serialization and is reflected in app-server
model output and structured telemetry.
GitOrigin-RevId: d24af30c3fc5820521b4beba1f9970714dad6482
## What changed
- Cover object serialization through the `text()` helper directly in the code
mode runtime suite.
- Remove redundant core integration coverage for serialized text and rejected
image outputs.
GitOrigin-RevId: f465b8361e9d725a7246994d9c41e9d01c969011
## What changed
Replace the core integration test for allowed `globalThis` properties with an
in-process code-mode runtime test. The test continues to fail when the runtime
exposes a global outside the allowlist without requiring the core network test
harness.
GitOrigin-RevId: f33846ae2bb35e6779b5bec0a3f0ba5729eb707d
## What changed
Move circular-value coverage for the `text()` helper from the core integration
suite to the in-process code mode runtime tests. Verify that stringification
returns no content and surfaces the V8 circular-structure error without needing
a mock server or network access.
GitOrigin-RevId: d506591af5ae8ce0a8c7305d633a7b0f2e458335
## What changed
- Add `create_session_with_limits` and session-scoped cell execution limits.
- Clamp execute and wait yield times to the session's `max_yield_time_ms`
without terminating the running cell.
- Negotiate support with remote code-mode hosts and include non-default limits
in `session/open`, while keeping unlimited sessions compatible with hosts and
providers that do not support limits.
## Testing
- Cover yield-time clamping, zero-timeout behavior, and isolation between
sessions.
- Cover wire serialization, capability negotiation, unsupported hosts, and
shared process-host execution.
GitOrigin-RevId: 9517321cd605bb87f93eeaa6ba331cc2e346e582
## Why
Code mode must link against a V8 build with sandbox support, but Windows MSVC
still used upstream non-sandboxed prebuilts and package builds selected the
older release artifact profile.
## What changed
- Enable the `v8_enable_sandbox` feature directly for the code mode runtime.
- Select the `ptrcomp_sandbox_release` archive and bindings in Cargo packaging
and CI, including Windows MSVC release builds.
- Point Bazel's Windows MSVC targets at the sandbox-enabled Codex artifacts and
enable the matching crate feature.
## Testing
Add a runtime test that calls `v8__V8__IsSandboxEnabled()` and verifies that
the linked V8 library has sandbox support enabled.
GitOrigin-RevId: c1b49b44a6ccfea5b5006d69ec7866848d1cddd7
## What changed
- Move the V8 implementation into a dedicated `codex-code-mode-runtime` crate used by `codex-code-mode-host`, removing the embedded runtime fallback from the Codex process.
- Resolve the host executable from the active installation layout and check its availability before selecting tools.
- Fall back to direct tools with a one-time warning when optional code mode is unavailable. Keep `code_mode_only` and `disable_in_process_fallback` configurations fail-closed.
## Testing
- Cover host discovery for standalone and package layouts, including missing hosts and symlinks.
- Verify direct-tool fallback, one-time warnings, and fail-closed code-mode-only behavior.
GitOrigin-RevId: 5aa3c6f1db148b2231fc24089a2ee0e2b00dbddb
Sean Huang