Who contributed to OpenAI on May 13, 2026?

4 developers shipped this update, including etraut-openai, apanasenko-oai, pakrym-oai, and jif-oai.

What were the notable OpenAI updates?

Refactor chatwidget input flow into modules, feat(exec-server): use protobuf relay frames, and code-mode: carry nested tool kind through runtime.

OpenAI

Codex, the SDKs, and the engine behind ChatGPT

Pick a date

Topics: AI / ML Full archive →

The Wire · Showcase

CODEX UNPACKS FIVE-PHASE REFACTOR AS EXECUTION LAYER TIGHTENS PROTOCOL

By RepoJournal · Filed 06:02 UTC on May 13, 2026 · About OpenAI

4 people shipped this

pakrym-oai @pakrym-oai 2 cited

etraut-openai @etraut-openai 1 cited

apanasenko-oai @apanasenko-oai 1 cited

jif-oai @jif-oai 1 cited

Codex is systematically decomposing its monolithic chatwidget while simultaneously hardening the remote execution contract with protobuf relay frames—two architectural moves that signal a shift toward cleaner boundaries between composition, execution, and protocol.

Phase 2 of the chatwidget refactor [1] extracted the input and submission flow into a dedicated module, continuing the effort to move coherent behavior domains out of a bloated single file. In parallel, the exec-server layer [2] moved to protobuf relay frames to serve multiple harness JSON-RPC sessions from one executor websocket, making ownership boundaries explicit: harness and executor endpoints now own sequencing, acks, and reassembly while rendezvous only routes frames. Code mode [3] tightened its own coupling by threading tool kind metadata through the runtime instead of re-looking it up from the router on execution, eliminating redundant spec lookups that kept execution path tangled. The dispatch layer [4] simplified by removing the `is_mutating` gate—a second concurrency policy that duplicated what `supports_parallel_tool_calls` already expressed, reducing session plumbing overhead. A dependency lock update [5] keeps the Python toolchain aligned.

Action items

→ Review chatwidget phase 2 refactor [ref:1] to understand the five-phase decomposition plan before next feature work openai/codex [plan]
→ Verify exec-server relay frame adoption [ref:2] in your harness integrations—ownership boundaries changed openai/codex [monitor]
→ Test code mode with nested tools [ref:3] to confirm tool kind threading works in your use case openai/codex [monitor]

References

[1] Refactor chatwidget input flow into modules ↗ openai/codex
[2] feat(exec-server): use protobuf relay frames ↗ openai/codex
[3] code-mode: carry nested tool kind through runtime ↗ openai/codex
[4] tools: remove is_mutating dispatch gating ↗ openai/codex
[5] fix: uv lock ↗ openai/codex

Quick answers

What shipped in OpenAI on May 13, 2026?: Codex is systematically decomposing its monolithic chatwidget while simultaneously hardening the remote execution contract with protobuf relay frames—two architectural moves that signal a shift toward cleaner boundaries between composition, execution, and protocol. In total, 33 commits, 33 pull requests, and 1 releases landed.
Who contributed to OpenAI on May 13, 2026?: 4 developers shipped this update, including etraut-openai, apanasenko-oai, pakrym-oai, and jif-oai.
What were the notable OpenAI updates?: Refactor chatwidget input flow into modules, feat(exec-server): use protobuf relay frames, and code-mode: carry nested tool kind through runtime.

CODEX HARDENS MCP RUNTIME STABILITY ACROSS FIVE CRITICAL FIXES

OpenAI shipped a coordinated hardening of the Model Context Protocol layer across Codex, patching race conditions in thread startup, resource access, and authority management that could leave stale configurations in production.

ai 66 shipped 1-min read

OpenAI 1 day ago

SDK RELEASES LOCK IN PROMPT CACHING AND SPEND CONTROLS

Both Node and Python clients shipped v6.49.0 and v2.48.0 overnight with admin spend limit APIs and relaxed prompt cache validation.

ai 103 shipped 1-min read

OpenAI 2 days ago

CODEX CONSOLIDATES THREAD STARTUP, NODE SDK ADDS REALTIME AND STANDARD SCHEMA SUPPORT

Codex refactored its threading and database layer for consistency while the Node SDK ships four new schema helpers and fixes audio streaming in a single release push.

ai 120 shipped 1-min read

OpenAI 3 days ago

CODEX FIXES WINDOWS EXECUTION BUGS WHILE TIGHTENING SECURITY REVIEW GATES

Codex shipped critical fixes for Windows process management and terminal navigation that were silently breaking execution sessions on Windows machines.

ai 104 shipped 1-min read

Elsewhere on the wire

AI Agents about 9 hours ago

CLAUDE OPUS 5 LANDS ACROSS THE STACK

The newest Anthropic model is now live in langchain, Cline, and llama-index, with native support for extended reasoning and 1M context windows.

ai-agents 28 shipped 1-min read

Local LLMs about 9 hours ago

OLLAMA LANDS LAGUNA SUPPORT AND CRUSHES MEMORY LEAKS WHILE SGLANG HITS V0.5.16 WITH CONFIDENCE-DRIVEN SPECULATIVE DECODING

Ollama shipped three critical performance and reliability fixes for Metal residency and concurrent access patterns, while SGL-Lang released 0.5.16 with a new speculative algorithm hitting 383.7 tok/s on DeepSeek-V4.