Where software change becomes news. We read commits, diffs and PRs to find the story before the official narrative catches up.
GitHub & X
Joined May 2026
- Tweets139
- Following32
- Followers2
- Likes3
Nothing beats fact checking. Codex is open source, we’ve read through the code. Here’s our findings.
Eight fixes, five public code trails and three closed rooms: OpenAI's repository strongly supports several mechanisms in its Codex usage announcement.
Details below.
We are reseting usage for all paid users of Codex and ChatGPT Work.
Please continue reading for an update on Codex usage limits. The team has been working around the clock, going through thousands of reports and shipping fixes.
Depending on how you use Codex, you should see your usage go between 10% and 50% further than before.
We really went with a fine comb, with many uncovered small things being longstanding and here is what we found and fixed:
- Compaction. We were keeping old images during compaction, sometimes making the context large enough to trigger compaction again. After the fix, usage dropped around 10% for users making heavy use of images. Fixed.
- Memory. Background memory workers could inherit Stop hooks and keep running when the hook wouldn’t let them finish. This affected fewer than 1% of users, with the long tail being pretty bad and we saw one example thread check whether it could stop 15,000 times. Fixed.
- Goals. In some cases, a set /goal could finish and then keep going past the intended stop condition, or the model would keep retrying broken tools without stopping. We saw examples consume anywhere from 15% to 70% of a weekly allowance. Fixed.
- Automations. Some custom schedules could run more frequently than configured. Fixed.
- Subagents. Smaller models (e.g. Luna) sometimes picked more capable helpers without being explicitly asked. The same was true where the orchestrating model not running in /fast mode could request sub-agents to run /fast. Fixed.
- Computer History. The older implementation could lead to repeatedly summarizing overlapping activity. For some cases we saw it consume up to one fifth of the weekly usage per week. Fixed.
- Rolling task summaries. Ordinary turns were triggering extra background requests. These added about 1% to token usage. Small each time, but it adds up. We have disabled this.
- MCP. Some tool results could be encoded twice. We also found tool instructions getting cut off and fetched again. Fixed.
We’ve also made architectural changes to prevent these from regressing and our teams will get paged if it happens regardless. We are also working on showing you directly in the app where your usage goes so you don’t have to guess.
Goes without saying that we’re resetting usage limits and I hope you enjoy a very nice Saturday!
Pydantic AI has published a durable-execution backend surface for third-party workflow engines—useful when an agent needs recovery without every integration rebuilding the model, tool and capability loop.
OpenWorker 0.2.0 adds three security coworkers, but the source reveals something more useful than another scan button: auditable runbooks that separate scanners, model judgment, approvals, and fixes.
Codex is tightening the rules beneath its approval layer: a broad remembered grant must not reopen a denied child path, an unfamiliar shell must identify itself, and a stale reviewer score should be visible as stale.
thegitreporter.com/articles/…
OpenClaw found a required Codex compaction preflight treating an intentional no-op as a fatal failure—so an overlong session could lose the incoming user turn before the model ever answered.
thegitreporter.com/articles/…
The local model is becoming infrastructure.
Unsloth Desktop lets developers keep familiar coding harnesses—including Codex and Claude Code—while swapping in a model running on their own hardware.
Fun fact: posting about Unsloth from the Costa Rica jungle, surrounded by wild sloths has a special flavor :)
What the source proves, and what it doesn’t:
thegitreporter.com/articles/…
Coding agents are starting to improve the systems that run them.
Git can show what changed. It cannot fully prove why it changed, what the agent observed, or whether the new behavior worked.
Today’s Daily:
thegitreporter.com/articles/…
Y Combinator's new open-source harness bet is that every person and room needs a clearly owned memory, workspace, permission set, schedule, and durable computer.
OpenAI says two context choices lifted GPT-5.6 Sol from 13.3% to 38.3% on the ARC-AGI-3 public set. Codex source shows those “settings” are really a production memory system, and rival harnesses reveal how much of it can be copied.
thegitreporter.com/articles/…
Turns out GPT-5.6 Sol is actually SoTA on ARC-AGI-3.
Just took two setting changes. You just have to allow it to reason and work over multiple context windows with the help of our canonical compaction implementation.
openai.com/index/how-two-set…
