# Codex’s Usage Reset Exposes Unbounded Agent Loops *By Coding Agents Alpha Tracker • August 30, 2026* A Codex audit found quota-burning failures in compaction, goals, automations, subagents, and MCP; the practical response is regression-testing agent control paths before unattended runs. ## 🔥 TOP SIGNAL **Codex’s usage reset exposes control-plane bugs in autonomous coding agents.** @thsottiaux says paid Codex and ChatGPT Work users are being reset after thousands of reports, with expected usage 10%–50% further depending on how Codex is used; the audit found retained-image compaction loops, `/goal` overruns or retries that consumed 15%–70% of a weekly allowance, over-frequent automations, unexpected helper escalation, and duplicated or truncated MCP results. [^1] Treat those as a regression suite: in a disposable repo, replay an image-heavy compaction, a `/goal` with a deliberate stop, a Stop-hook interaction, a custom schedule, and a large MCP result; log token deltas, helper selection, and whether the run terminates. This test plan is my recommendation, based on the failure modes the update says were fixed. [^1] ## ⚡ TRY THIS - **Turn idea dumps into a pull queue.** ThePrimeTime’s Linear MCP workflow is simple: finish a long brainstorm, have the agent turn the ideas into tickets, then—when you do not know what to work on—run: **“Hey, Agent, go to Linear and tell me what I need to do next.”** Resource: [Linear Terminal](https://linear.app/terminal). [^2] - **Make a side-effecting agent draft-only by construction.** Kent C. Dodds’s Kody pattern starts with the explicit grant **“Create a draft reply. Never send.”** Put that intent in the README and export docs; request `gmail.compose` plus `gmail.googleapis.com`, add `gmail.readonly` only if inbox reading is required, call `users.drafts.create` rather than a send endpoint, then publish a thin drafts-only package and lock it with `changes: { locked: true }`. Gmail’s OAuth scope can still technically send, so the package lock—not the token—is the authority boundary; later publishes require the owner’s **Promote this commit** action. [^3] - **Use real file paths for remote context.** In the T3 Code nightly, attach any file; the attachment is exposed to the agent as a real path. Upload a PDF, Markdown file, or MP3 and let the agent operate on that path instead of building a bespoke transfer step. [^4] - **A/B a cheap model with a deterministic smoke test.** Matthew Berman’s GLM workflow is replicable: create a Z.AI or OpenRouter API key, connect it to OpenCode—or another client supporting an OpenAI-compatible endpoint—and avoid sensitive data when using the China-served Z.AI endpoint. His baseline task was a Rubik’s Cube simulation; he checked scrambling, solving, cube size, colors, speed, camera, lighting, and materials. [^5] ## 📡 WHAT SHIPPED - **GLM 5.3 Flash (Z.AI) is the main model-level release signal.** Berman presents it as an open-weights mixture-of-experts model with 320B total parameters and 18B active, reporting Terminal Bench 84.3 and DeepSuite 63.4; he also cites a 1M-token context, 131K maximum output, and maximum reasoning by default. [^5] His cost estimate is about **$0.09 per intelligence-index task**, but the model uses about **47,000 output tokens** on average versus roughly 20,000 for Luna; he still calls Luna the better pure cost/quality trade-off, while GLM buys open-weight control. [^5] His same-prompt demos are not a controlled benchmark: GPT 5.6 Soul had internet access in Codex while OpenCode did not. [^5] - **Claude Code’s capacity schedule is changing on September 14.** ClaudeDevs says the current 50% weekly-limit increase remains in place until then, after which standard limits for Pro, Max, Team, and seat-based Enterprise plans will be permanently 25% higher than the prior standard; the team characterizes that as a 17% reduction from today’s promotional level. [^6][^7] Treat September 14 as the budget boundary for long-running agent work. - **T3 Code’s adoption signal is now substantial.** Theo reports more than **250,000 users** and **70,000 weekly active users** nine days after an earlier update, and says Linux became the most popular platform as of August 23. [^8][^9] Omarchy 4.1 is slated to ship with T3 Code, with DHH describing setup as taking only a few seconds. [^10] - **Provider exposure needs task-level accounting.** In the Cursor discussion, @thsottiaux cautions that the cited 5% model-traffic share is not a proxy for revenue or value because token-efficient frontier models consume fewer tokens; a current-period video separately quotes Harrison Chase arguing that a harness owned by no model lab is the only architecture that works across every model. [^11][^12] ## 🎬 GO DEEPER - **Matthew Berman — [“Cancel your subscriptions, Ox-Alpha is here! (GLM 5.3 Flash)”](https://www.youtube.com/watch?v=TOWXXhn7ctY):** Watch the API hookup, Rubik’s Cube smoke test, and same-prompt comparison. The useful lesson is how to combine a deterministic artifact test with cost, token-efficiency, and tool-access caveats rather than trusting a leaderboard claim alone. [^5] [![Cancel your subscriptions, Ox-Alpha is here! (GLM 5.3 Flash)](https://img.youtube.com/vi/TOWXXhn7ctY/hqdefault.jpg)](https://youtube.com/watch?v=TOWXXhn7ctY&t=679) *Cancel your subscriptions, Ox-Alpha is here! (GLM 5.3 Flash) (11:19)* - **OpenAI Sets a Date to Cut Cursor Off:** Skip the ownership drama and watch the Amazon/Kiro workflow section. The video reports that, across roughly 50 Amazon Stores teams, merely adding AI produced under 3× prior deployment velocity while teams that changed how they worked reached 4.5× or better, with documentation, precise errors, compiler feedback, and locally mocked services offered as the practical levers. The figures are self-reported and deployment velocity is not the same as value shipped. [^12] [![OpenAI Sets a Date to Cut Cursor Off](https://img.youtube.com/vi/yJxgxSIlJI0/hqdefault.jpg)](https://youtube.com/watch?v=yJxgxSIlJI0&t=620) *OpenAI Sets a Date to Cut Cursor Off (10:20)* - **Study [Kody’s locked Gmail drafts guide](https://kody.codes/guides/locked-gmail-drafts):** It is a compact example of separating authentication from declared authority: a published export creates a draft and returns its ID, while the human reviews and sends; the lock prevents this package’s jobs and exports from silently becoming a sender. [^3] **Editorial take:** The highest-alpha agent work today is not adding autonomy; it is making autonomy observable, interruptible, and narrowly authorized—usage accounting, deterministic smoke tests, and locked side effects are becoming first-class features. [^1][^3] --- ### Sources [^1]: [𝕏 post by @thsottiaux](https://x.com/thsottiaux/status/2093801758665715784) [^2]: [1968 Predicted Every Modern Tech Problem | TheStandup](https://www.youtube.com/watch?v=gbzLuAzhJ0c) [^3]: [Gmail drafts without send — lock what Google cannot scope](https://kody.codes/guides/locked-gmail-drafts) [^4]: [𝕏 post by @theo](https://x.com/theo/status/2093622750510563606) [^5]: [Cancel your subscriptions, Ox-Alpha is here! \(GLM 5.3 Flash\)](https://www.youtube.com/watch?v=TOWXXhn7ctY) [^6]: [𝕏 post by @ClaudeDevs](https://x.com/ClaudeDevs/status/2093742321473065266) [^7]: [𝕏 post by @ClaudeDevs](https://x.com/ClaudeDevs/status/2093742322525810912) [^8]: [𝕏 post by @theo](https://x.com/theo/status/2093844595268522430) [^9]: [𝕏 post by @theo](https://x.com/theo/status/2093872171584213318) [^10]: [𝕏 post by @dhh](https://x.com/dhh/status/2090124335642038766) [^11]: [𝕏 post by @thsottiaux](https://x.com/thsottiaux/status/2093784314714657110) [^12]: [OpenAI Sets a Date to Cut Cursor Off](https://www.youtube.com/watch?v=yJxgxSIlJI0)