RepoJournal
OpenAI OpenAI
150 wires and counting

$ follow OpenAI

Keep up with OpenAI in about 3 minutes: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

fair warning: these emails are deeply technical. diffs, version numbers, CVEs, benchmark deltas. if that's not your idea of a good read, this isn't your newsletter.

Folds into your digest — weekly by default, monthly if you prefer. Unsubscribe in one click.

$ status

wire 2026-06-28
stories 13
RSS Archive Reviews

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

CODEX STABILIZES PROMPT CACHE WITH SYNTHESIZED OUTPUT IDS

By RepoJournal · Filed · About OpenAI · Composed from the cited sources · methodology

xl-openai maxj-oai nhamidi-oai
3 people shipped this
xl-openai @xl-openai 1 cited maxj-oai @maxj-oai 1 cited nhamidi-oai @nhamidi-oai 1 cited

Codex is fixing a subtle bug that was breaking prompt-cache reuse on retries and resumes by assigning stable IDs to synthesized outputs.

The core issue: when `ContextManager::for_prompt` repairs an unmatched call by synthesizing an aborted output, that output was getting a fresh ID on every prompt rebuild, which meant identical model context would hash differently and waste cached prompts [1]. That's now fixed with stable ID assignment, letting retries and resumes actually hit the cache. In parallel, Codex is enforcing marketplace source policy at runtime [2], filtering plugin discovery and CLI reporting through enterprise policy so blocked plugins become inactive. The app-server is also exposing a new environment-info RPC [3] so clients can discover shell and working directory before selecting an execution environment, with bounded response times to prevent hangs. Finally, custom tool calls now preserve their optional namespace through response deserialization and app-server replay [4], fixing routing and streaming argument handling for namespaced tools.

$ ls openai/reviews/ # the step-back reads

the week in review Codex hands model defaults to the server the month in review Agents API lands in both OpenAI SDKs
Full archive → Weekly reviews →

Quick answers

What shipped in OpenAI on June 28, 2026?
Codex is fixing a subtle bug that was breaking prompt-cache reuse on retries and resumes by assigning stable IDs to synthesized outputs. In total, 5 commits, 5 pull requests, and 3 releases landed.
Who contributed to OpenAI on June 28, 2026?
3 developers shipped this update, including xl-openai, maxj-oai, and nhamidi-oai.
What were the notable OpenAI updates?
core: stabilize synthesized call output IDs (#30327), [plugins] Enforce marketplace source policy at runtime, and [app-server] expose environment info RPC.
♥

$ log OpenAI --recent

Safety identifiers reach images, plugin approvals bind to contents 93 shipped Credential masking lands in Codex, Live gains Decisions across both SDKs 100 shipped Codex stabilizes environment tools, python adds residency options 135 shipped openai-python restores streaming connection reuse, browser and computer toolsets land in both SDKs 123 shipped

$ log --all # elsewhere on the wire

Alignment & Post-training verl rejects conflicting vLLM engine_kwargs, open-instruct adds MILES OPD behind a Megatron learner Interpretability nnsight 0.8.0 cuts import time from 7.6s to 1.6s

$ ls showcases/

Safety Evals AI Red Teaming Agent Security Interpretability Alignment & Post-training

$ init your-desk

RepoJournal can write one of these for any repo you care about. Sign in with GitHub, pick your repos and sources, and your desk gets a wire like this one every morning.