
https://nitter.cf/t.co/9Od4BYinYs https://nitter.cf/t.co/o05daXPcOQ https://nitter.cf/t.co/c4l6UiFWMD Collab: [email protected]
London, England
Joined March 2022
- Tweets6.2K
- Following2.8K
- Followers12.2K
- Likes19.1K
Opening a managed Browser Run session is not the same as running your own Kitesurf engine.
Cloudflare's Kitesurf is an experimental browser engine for AI agents. Official changelog, 12 November 2026: the engine is now open source under Apache 2.0. It parses HTML, applies CSS, runs page JavaScript, and renders screenshots on Cloudflare Workers, without a Chrome container. It speaks the Chrome DevTools Protocol, so clients such as Puppeteer can connect.
Official README requirements before you build: Bazelisk installed as bazel, Node.js 22 or newer, and pnpm 9.15.4, the version pinned in the repo. On macOS, GNU tar is also required.
Then:
git clone github.com/cloudflare/kitesu…
cd kitesurf
pnpm install --frozen-lockfile
pnpm run dev
The first build compiles the Rust core, so it can take a while. Official docs: the engine listens at http://127.0.0.1:8787. Visit http://127.0.0.1:8787/health to confirm it is running. To drive it from Puppeteer, official examples connect puppeteer-core to ws://127.0.0.1:8787/devtools/browser, open a page, and write a screenshot. I have not run this build.
If you later deploy your own instance, that path is different. Official warning: Cloudflare deployment requires a Workers Paid plan because Kitesurf uses Dynamic Workers to isolate pages. Local wrangler dev does not. Official README: do not deploy without authentication in front. The engine has no auth of its own. A person who has the URL can open DevTools, connect a CDP client, send outbound requests from your account, and consume CPU time and Dynamic Workers that you pay for. The workers.dev URL is public by default and follows kitesurf.<your-subdomain>.workers.dev.
Official Browser Run docs still list work the engine cannot do yet: play video or render WebGL, negotiate a bot-challenge handshake with real TLS fingerprints, or start a long-running authenticated session that needs persistent state. For those cases, official docs point to Browser Run's default Chromium browser.
Changelog: developers.cloudflare.com/ch…
Blog: blog.cloudflare.com/kitesurf…
Repo: github.com/cloudflare/kitesu…
#AI #AIagents
2
198
A spending cap in Cloudflare is not on by default, and an email alert is not a cap.
Cloudflare Spend Management, documented on 13 November 2026, lets you set a spending limit or an email alert for Workers and Durable Objects. Official docs: more products are coming, but those two already have separate account-wide controls. A Workers limit does not include Durable Objects, Workers AI, Workers KV, R2, or other products your app may use. You also cannot target one Worker, one Durable Object namespace, or one billing metric.
If you want a hard stop, create a usage_block policy. Official example, a $50 Workers limit for the current billing period:
cf billing spend-policies create \
--name 'Workers $50 spending limit' \
--product-id workers \
--enabled true \
--threshold-cents 5000 \
--action usage_block
5000 is $50 in US cents. For Durable Objects, change the product id and the name.
Three checks after you save it:
1. Confirm the policy is On. Saving a new limit turns it on. Official warning: the limit counts usage already recorded this period, so saving $50 after the product has already accrued $60 can pause it as soon as Cloudflare evaluates the policy.
2. An email alert is a different control. Official docs: alerts notify you and the product keeps running. Dashboard-created spending limits do not automatically get an email recipient; add notification_email if you want those messages. Alerts fire at 80% and 100% of the threshold.
3. A limit does not guarantee a maximum invoice. Metering delays, in-progress operations, and enforcement failures can let spend exceed it. For Durable Objects, stored data can still incur charges while requests are paused. The Workers Paid subscription fee is billed separately. The period follows your subscription renewal date, not the first of the calendar month.
You need permission to edit billing settings. I have not created a policy on an account.
Docs: developers.cloudflare.com/bi…
Changelog: developers.cloudflare.com/ch…
Cloudflare's post: nitter.cf/Cloudflare/status/2122…
#AI
Spend Management adds opt-in spending caps for Cloudflare Workers and Durable Objects, and more products soon. Here is how we built the metering pipeline that turns usage into runtime decisions. cfl.re/4iX4BRe
1
146
Installing GitHub's security extension does not automatically show your repository's code scanning alerts.
The plugin is in public preview for the GitHub Copilot app. Official docs split two jobs.
Install it with:
copilot plugin install github-security@awesome-copilot
Or in the app: Customize, search github-security, then Install. Official docs: it enables automatically after install.
Then check which path you actually have:
1. Local review of current changes. In a session with in-progress edits, open Canvas > Security and start a review, or type /security-review. Official docs: this works on any Copilot plan. You do not need code scanning or a GitHub Code Security license. It is a review of your session changes, not a full repository audit, and it does not enable code scanning.
2. Existing code scanning alerts. Those appear in the Security canvas only if code scanning is enabled and you can access the alerts. Private repositories also need GitHub Code Security. The canvas shows code scanning alerts, not Dependabot or secret scanning alerts.
If you click Fix, official docs say Copilot changes only your local working copy. It does not commit, push, or open a pull request, and a local fix does not close the alert until the change reaches the branch and code scanning runs again. Dismissing an alert does update it on GitHub. If your organization uses delegated dismissal, the extension submits a request and the alert stays open until someone approves it.
Explain, Fix, and /security-review use the model selected in the app and consume AI credits. Browsing and dismissing alerts do not. Business and Enterprise need the GitHub Copilot app policy enabled. GitHub Enterprise Server is not supported.
I have not installed the plugin or run a review.
Docs: docs.github.com/en/copilot/h…
Changelog: github.blog/changelog/2026-1…
GitHub's post: nitter.cf/github/status/21220585…
#AI
🔒 Check your repo's code scanning alerts without leaving the GitHub Copilot app.
With the new GitHub security extension, you can ask Copilot to find, explain, and fix (or dismiss!) an alert.
Now in public preview ⬇️
docs.github.com/copilot/how-…
2
1
200
DY retweeted
Extract the design system first. Applying it to screens is a second official skill.
Google Labs published Stitch Design Skills, a library of agent skills for Google Stitch, compatible with Codex, Antigravity, Gemini CLI, Claude Code, Cursor, and a manual OpenCode path. Official README: these skills need the Stitch MCP server configured and running in the agent environment. Installing the skill files is not enough if that server is missing. Official MCP setup: stitch.withgoogle.com/docs/m…
Official docs also warn that the skills often depend on each other. If you install only some of them, include the required dependencies.
Two official design skills, in order, with their documented prompts:
1. extract-design-md. "Scan /src and extract the design system into .stitch/DESIGN.md."
2. manage-design-system. "Upload our design system from .stitch/DESIGN.md and apply it to all screens."
That gives you a file you can open and review before anything is applied. I have not run these skills or Stitch MCP.
To browse or install from the official repo:
npx skills add google-labs-code/stitch-skills
Claude Code, Cursor, Codex, and OpenCode have different official install paths in the README, including plugin marketplaces and, for OpenCode, a manual copy plus a remote MCP entry. Follow the path for the client you actually use.
Repo: github.com/google-labs-code/…
#AI
A 3D mockup does not have to become a flattened image.
Figma's new 3D transforms keep the underlying layers editable. Try this small exploded-card composition instead of asking for “something futuristic.”
Starting recipe:
Parent frame: 720 × 480
Three card frames: 240 × 160
Labels: Plan / Build / Ship
Positions: x=40, 240, 440; y=160
Each card: Rotate X=8°, Y=-18°, Z=0°
Depth: 0 / 40 / 80
1. Select each card. Effects > + > 3D transform, then enter its rotation and Depth.
2. Select the parent. Enable Keep nested layers 3D, then add Perspective and start at 800px.
3. Edit the word “Plan” after applying the effect. It should remain text, not pixels in a replacement image.
Optional agent prompt:
“Build this three-card composition using editable text and frames. Use the dimensions and transforms above. Keep each card separate and preserve the shared 3D space. Do not replace the composition with a generated image.”
Treat these numbers as a design starting point, not a render I tested. Official docs: smaller perspective distances exaggerate depth; larger ones make it subtler. Compare both before choosing.
Open beta. Full seat and edit access required, even though the feature is on all plans.
Demo: nitter.cf/figma/status/212092666…
Official steps: help.figma.com/hc/en-us/arti…
3D TRANSFORMS ARE HERE
→ Explore dimension, perspective, and depth with x, y, and z rotation
→ Build and animate 3D designs with help from Figma agent
→ Implement with your coding agent via Figma MCP
Rolling out today in open beta
196
One messy note. A seven-field brief. A Workspace skill you can reuse.
Google is rolling out reusable Workspace skills. Here is a complete instruction to try with a current document:
Skill: Brief Builder
“Turn the current document into a work brief with these fields: objective, audience, deliverable, duration, required assets, deadline and decisions needed. Use only stated facts. Mark absent facts as Unknown. Keep conflicting requirements separate. Finish with the single question that would most unblock the work. Do not invent a date, budget or owner.”
Synthetic input:
“We need a 30-minute onboarding workshop for new hires. Include an accessibility checklist. The date is undecided.”
Expected fields to check:
Objective: onboarding
Audience: new hires
Deliverable: workshop
Duration: 30 minutes
Required asset: accessibility checklist
Deadline: Unknown
Decision needed: date
Question: What date should the workshop happen?
In Docs, open Ask Gemini with skills > Tools > Skill builder, paste the instruction, review it, then choose Make this skill in Studio and Turn on. Generating the instructions alone does not save an enabled skill. Use the enabled skill through the app's skill mention picker.
Recipe, not my account test. English-only, eligible plans and admin controls apply. Scheduled Release starts November 30.
Google's announcement: nitter.cf/GoogleWorkspace/status…
Setup: support.google.com/workspace…
Every team member brings their own expertise and know-how to work — the invoicing expert or your colleague who knows where to find every design asset.
You can now scale up that institutional knowledge with new Workspace skills, so everyone can do fewer repetitive tasks and bring best practices to their work. Read more 👉 goo.gle/3USossu
1
268
A countdown timer that looks like studio hardware, not another default web form.
Shu Ding's SPEC-01 is a hardware-themed component kit built on Vercel UI SDK. The useful starting point is a complete device, not a screenshot to imitate.
Try the official countdown timer in a disposable React project already initialized for shadcn, using the default @/components aliases:
npx shadcn@latest add spec-01.vercel.app/r/timer.j…
Render it in a client component:
"use client"
import { Hardware } from "@/components/ui/hardware"
import { Timer } from "@/components/blocks/timer/timer"
export default function App() {
return (
<Hardware keyColor={{ l: 0.64, c: 0.31, h: 48 }}>
<Timer />
</Hardware>
)
}
The registry source includes the timer, its hardware skin and dependency links. Hardware scopes the look to its children, so you can try one panel rather than restyle the whole app.
Acceptance check: change the duration, Start, Pause, then Reset. Check the actual countdown state, not just the nice dial. The source also includes a chime toggle; browser audio still needs testing in your app.
Source-reviewed recipe, not a claim I installed or tested it.
Original demo: nitter.cf/shuding/status/2121609…
Kit: spec-01.vercel.app
CLI: ui.shadcn.com/docs/cli
1
2
336
A launch decision board, corrected before the call ends. Not another unread recap.
Claude Meetings can build a whiteboard during an internal Google Meet call, not just summarize it afterward. Here is a live launch-choice exercise to try.
Say this during the call:
“Claude, make a whiteboard titled Launch choice. Put Friday and Monday side by side. Under each, show what is ready, blockers and evidence we still need, using only this meeting. Add a decision strip underneath with chosen option, reason and next action. Leave it undecided until we explicitly agree. Update the board when we correct it.”
Two synthetic checks:
“Friday would be nice” should stay an option, not become a commitment.
“We are choosing Monday. Maya owns mobile QA” should update the choice and action owner. A deadline that nobody stated should remain unconfirmed.
Before ending the call, open the meeting-page link and have the team correct the actual board. That is the deliverable: a shared decision artifact, not another paragraph nobody reviews.
Official setup: Meetings > connect Google Calendar > select the call > Add Claude. Your organization must have the Calendar connector enabled. Make sure everyone agrees to transcription before adding it.
Untested workflow recipe. Rolling out in beta on Team and Enterprise; some organization configurations are excluded. Keep this exercise to an internal team call; outside guests do not get the Claude meeting page.
Original demo: nitter.cf/claudeai/status/212097…
Guide: support.claude.com/en/articl…
Introducing Claude Meetings: add Claude to your team’s video call and it builds docs, whiteboards, and other artifacts with you while you’re still talking.
Starting to roll out in beta on Team and Enterprise plans.
1
1
249
Extract the design system first. Applying it to screens is a second official skill.
Google Labs published Stitch Design Skills, a library of agent skills for Google Stitch, compatible with Codex, Antigravity, Gemini CLI, Claude Code, Cursor, and a manual OpenCode path. Official README: these skills need the Stitch MCP server configured and running in the agent environment. Installing the skill files is not enough if that server is missing. Official MCP setup: stitch.withgoogle.com/docs/m…
Official docs also warn that the skills often depend on each other. If you install only some of them, include the required dependencies.
Two official design skills, in order, with their documented prompts:
1. extract-design-md. "Scan /src and extract the design system into .stitch/DESIGN.md."
2. manage-design-system. "Upload our design system from .stitch/DESIGN.md and apply it to all screens."
That gives you a file you can open and review before anything is applied. I have not run these skills or Stitch MCP.
To browse or install from the official repo:
npx skills add google-labs-code/stitch-skills
Claude Code, Cursor, Codex, and OpenCode have different official install paths in the README, including plugin marketplaces and, for OpenCode, a manual copy plus a remote MCP entry. Follow the path for the client you actually use.
Repo: github.com/google-labs-code/…
#AI
Three tidy spreadsheet rows. One double-booked room.
A tiny test case for Google's new Sheets canvas. Copy this synthetic one-day agenda into a Google Sheet:
IDRoomStartEnd
ARoom A09:0009:30
BRoom A09:2009:50
CRoom B09:2009:40
On a computer, open Insert > Create a canvas. Try this prompt:
“Build a one-day run-of-show with a lane per room and draggable session blocks. Use Start and End as 24-hour local times. Keep IDs stable. Flag overlapping sessions in the same room without moving them automatically. When I drag a block, preserve its duration and update the backing Start and End cells.”
Three acceptance checks:
1. A and B overlap for 10 minutes. C is in another room, so it is not a conflict.
2. Move B to 09:30. Its new end should be 10:00.
3. Open View data and check the actual cells changed. A prettier screen alone is not a pass.
This is a proposed test recipe, not a claim I ran the generated canvas. The official demo shows a draggable agenda synced to its sheet; the overlap behavior above is what to ask for and verify.
Availability depends on account eligibility; English only. Use a private copy of sample data.
Google's demo: nitter.cf/GoogleWorkspace/status…
Setup: support.google.com/docs/answ…
Planning a multi-track conference or launch?
Map your event run-of-show onto an interactive visual timeline with Sheets canvas. Spot gaps, adjust session timings with simple drag-and-drop, or add new agenda items on the fly. Any changes you make are synced back to the underlying sheet. goo.gle/4yqbmkl
3
279
Installing the Herdr skill does not put your agent inside Herdr.
Official Herdr docs split two files. The official skill file teaches an agent how to control Herdr from inside a Herdr-managed pane: inspect workspaces, tabs, and neighboring agents; split panes and run commands without stealing focus; read pane output; wait for servers, tests, or another agent; and start helper agents in sibling panes. A separate official guide is for an agent helping a human learn, set up, or troubleshoot Herdr. Those are different jobs. The skill file does not install Herdr itself.
Install the skill with:
npx skills add herdrdev/herdr --skill herdr -g
Omit -g to install it only in the current project. If you previously installed it when the skill lived at the repository root, official docs say to rerun this add command once rather than skills update.
The skill is only an instruction file. After it is installed, start Herdr, then start the coding agent in a Herdr pane so HERDR_ENV=1 is set. Official skill rule: if that variable is not set, the agent should stop and say it is not running inside a Herdr-managed pane. That is there so an agent outside Herdr does not try to control a session it does not own.
If Herdr is already installed, herdr --skill prints the copy bundled with that binary, so the command reference matches the release you are running. Official source of truth: the repository copy of the skill file.
I have not run Herdr myself.
Docs: herdr.dev/tui/docs/agent-ski…
Skill file: github.com/herdrdev/herdr/bl…
#AI #AIagents
1
244
Write the spec, then stop. This official skill is not supposed to start the plan or the code in the same turn.
Addy Osmani's spec-driven-development skill, in the MIT-licensed agent-skills repo, is for a new feature or project when no specification exists yet, or when requirements are still vague. Official docs say to skip it for a one-line fix, typo, or a change whose requirements are already unambiguous.
The gated workflow is Specify, then Plan, then Tasks, then Implement. A human reviews each phase before the next one starts. If one request actually bundles several independently testable capabilities, official docs add a Phase 0 capability map: stable module ids, one-way dependencies, and a build order, reviewed before any module spec is written.
A single-capability spec is supposed to cover six areas: objective, commands (full executable commands, not just tool names), project structure, a real code-style example, testing strategy, and boundaries split into Always, Ask first, and Never. It should also include testable success criteria and open questions. Official instruction: list assumptions before writing the spec, and wait for a later-turn approval after the spec file is saved.
To install only this skill:
npx skills add addyosmani/agent-skills --skill spec-driven-development
Official README caveat: a per-skill install copies that skill folder only, not the repo-level references directory, so shared checklists may be missing. The skill still works. I have not run this skill.
This is a source-linked AgentSkillsDir entry, not Addy Osmani's official site, and not a review or safety certification. The listing is pinned to commit aeac41d (upstream 13 October 2026). The skills CLI installs the current default branch, which may be newer than that pin. The directory's last star figure, 108,232 on 19 October 2026, is stale; I did not get a live GitHub count this morning.
agentskillsdir.com/skills/ad…
github.com/addyosmani/agent-…
#AI #AIagents
Before comparing two AI reviewers, compare what they actually received.
A reader described reviewers rejecting a true statement after their checker received only the first 900 characters. Here is a tiny input-plumbing reproduction, not a model benchmark:
from hashlib import sha256
source = "Background.\n" * 100 + "Decision: release is blocked.\n"
views = {"A": source[:900], "B": source}
for name, text in views.items():
data = text.encode("utf-8")
print(name, len(data), sha256(data).hexdigest()[:12], "Decision:" in text)
Output from a local run with synthetic text:
A 900 21e2b7901b8c False
B 1230 2756ab4e3afc True
A never received the decision line. B did. The source document was the same; the delivered bytes were not.
Before calling it model disagreement, save the exact delivered text, byte length, full SHA-256 and source version for each review. Keep sensitive text private.
Different hashes prove different bytes, not which verdict is correct. Matching hashes alone do not prove the instructions, hidden context or evidence were sufficient.
Reader's case: nitter.cf/shaharfrish/status/212…
Python hashing: docs.python.org/3/library/ha…
Replying to @DYoung_AI
A person first, then. Reviewers can share an input too: two of mine refused the same true statement three rounds running. The checker had sent them only the first 900 characters of a long source file. Do you log what each labeler saw on disputed cases?
1
1
307
The strategy was human. GPT-6 Astra led the detailed proof. The result is still a candidate.
OpenAI's science blog, dated 15 November 2026, published its first guest post this morning. Physicist Nathaniel Craig, a visiting scientist at OpenAI who leads the particle theory group at UC Santa Barbara, describes an October preprint prepared with GPT-6 Astra: a candidate nonperturbative lattice definition of the Standard Model.
Official post: Craig already had a strategy, but had not been able to prove it worked in every case it needed to cover. Astra helped turn that strategy into a construction, suggested how to carry force fields onto finer grids, showed that those comparisons would converge, and took the lead in developing the detailed proof. Lattice field theorists Maarten Golterman, Andreas Kronfeld, and Yigal Shamir reviewed Astra's proof and asked questions as the construction took shape.
Official caveats, in Craig's own framing: this is a beginning, not an end. Expert dialogue still has to decide whether the prescription has the properties the field wants. It is far more an answer in principle than a method ready for realistic lattice simulations. He is still examining whether the construction satisfies reflection positivity at finite lattice spacing. The interacting theory in continuous spacetime remains open.
I have not reviewed the preprint or checked the proof myself. This is the official collaboration log, not a claim that the Standard Model is now fully defined.
Blog: openai.com/index/step-toward…
OpenAI's post: nitter.cf/OpenAI/status/21221234…
#AI
Physicist Nathaniel Craig shares how he worked with GPT-6 Astra to develop a candidate lattice definition of the Standard Model.
It’s the first post on our new science blog:
openai.com/index/step-toward…
1
1
327
Turning on Vercel Security Review can cover repositories that are not linked to a Vercel project.
Official Vercel changelog, 13 November 2026: Security Review can scan GitHub repositories that have not yet been linked to a Vercel project. By default, it covers all unlinked repositories the Vercel GitHub App can access in a connected GitHub account. You can limit coverage to selected connected accounts or individual repositories in Agent settings. Once enabled, each push to a covered repository automatically triggers a review of its code changes.
That default is the part to check before you leave it on. Official docs: every push to a project or repository you select starts a billed review. With All repositories, that includes every repository the Vercel GitHub App can access in your connected GitHub accounts. Repository reviews and project reviews are billed the same. To stop reviews of repositories without a project, select No repositories under Repositories to Review. To stop all reviews, turn off Review Changes Automatically.
Official docs also split the two review types. A project review needs a Vercel project connected to the GitHub repository, is scoped to the project's root directory, and can use team and project policies. A repository review needs the repository's GitHub account connected to your team, covers the whole repository, and uses team policy only. Findings appear under the repository in the team's Vulnerabilities dashboard, not under a project.
Reviews start from push events, not only pull requests. Official docs: direct pushes and force pushes to your production branch are covered too.
On a public repository, official docs say Security Review posts only its check and keeps findings in the dashboard, so an unpatched vulnerability is not disclosed in the public pull request. On a private or internal repository, findings can be posted as pull request comments, which anyone with access to that repository can read.
Official docs: Security Review analyzes your code, not your running application. It is not a security audit, and it does not perform runtime or penetration testing.
I have not run Vercel Security Review myself.
Changelog: vercel.com/changelog/vercel-…
Docs: vercel.com/docs/agent/securi…
#AI
1
278
Code Mode is not the same Claude Code session with fewer tool trips.
Official Claude Code docs describe an experimental, off by default feature that may change or be removed in any release. With Code Mode on, Claude writes JavaScript that makes several tool calls in one step, then reads back only what the script returns. You see the tool calls, such as a diff. The script itself is behind Ctrl+O.
Official docs do not promise a token-savings percentage.
Before you turn it on, official limits and requirements:
1. It needs Claude Code v2.1.323 or later, the native install, and an interactive terminal session. It does not work with -p, the Agent SDK, the VS Code panel, the Desktop app, or cloud sessions.
2. It does not run through a gateway, Amazon Bedrock, Claude Platform on AWS, Google Cloud's Agent Platform, or Microsoft Foundry. HIPAA and zero data retention organizations are excluded. Official model list: Opus 5.5, Sonnet 5.5, Haiku 5.5, or Fable 5.2.
3. Remote Control, Claude in Chrome, and computer use are unavailable in a Code Mode session. You also cannot move the session to the Desktop app or turn it into a background session.
4. Resume only in the terminal, with Code Mode still on. If the variable is off, official docs say Claude Code refuses to resume that session.
5. In auto mode, the classifier reviews the script, not each tool call, and it does not see values computed at runtime, such as a path taken from search output. Keep deny or ask rules, or PreToolUse hooks, for calls you must control.
To try it: set CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1, confirm the first session prompt, and look for "Code Mode (experimental) is on". If it does not start, official docs say the reason appears on the same line as "Code Mode (experimental) isn't available", and /status shows it too. Official docs: a repository's project settings cannot turn this on for you. Organizations can set the same variable to 0 in managed settings to block it.
I have not run Code Mode myself.
Docs: code.claude.com/docs/en/code…
Release notes (v2.1.323, 12 November 2026): github.com/anthropics/claude…
#AI #AIagents
4
2
316
Codex on a Free or Go plan is not the same as Codex Cloud.
Official OpenAI help, under Availability: Codex is included across ChatGPT plans, including Free and Go. Codex Cloud is available to eligible Plus, all Pro tiers, Business, Enterprise, Healthcare, and Education accounts, subject to rollout and workspace settings. It is not included with Free or Go.
Official checks before you assume a feature is in your plan:
1. Sign in with your ChatGPT account in the client you actually use: the desktop app's Codex mode, Codex CLI, the IDE extension, or Codex web. Usage limits vary by plan; OpenAI points to chatgpt.com/pricing for higher-usage and credit options.
2. If the workflow you want is Codex Cloud, confirm you are on an eligible plan. Free and Go are not included.
3. Auto-review: when you are signed in with a ChatGPT account, official help says those safety checks are free and do not count toward your plan's usage limits. They do not remove your workspace's restrictions or replace every approval prompt. If the control is available, open the approvals menu in the ChatGPT desktop app composer and select Approve for me.
4. ChatGPT Work follows the same usage structure as Codex, per the Work and Codex help article. Work is a different product from Codex; this is about the usage structure, not a claim that the two do the same job.
I have not compared plan bills myself.
Codex and your plan: help.openai.com/en/articles/…
Work and Codex: help.openai.com/en/articles/…
#AI
2
4
464
A new Codex Cloud task does not bring back the other task's uncommitted files.
Official OpenAI help: each cloud task starts from your published environment in its own isolated workspace. You can return to the original task to continue unfinished work. Starting a new task creates a separate workspace and does not recover another task's uncommitted changes. Commit important work to source control.
The recovery window is easy to misread. Official help: by default, saved virtual machine state is recoverable for up to 7 days after the last start of a turn or task resume. That window describes saved VM state. It is not a conversation-history retention period.
Environment edits have the same split. Changes to a published environment apply to new tasks. Existing tasks keep their own state.
I have not run a Codex Cloud recovery myself. Official checks before you walk away from a task:
1. If the work matters, commit it to source control while you are still in that task.
2. If you need to continue, reopen the same task, not a fresh one.
3. If you change tools, dependencies, or access in the published environment, expect that to show up on the next new task, not the one already running.
4. Cloud tasks can continue while your computer is asleep, after the environment is created and published on desktop or web. Remote access to a task running on your own computer is a separate workflow.
Availability depends on your ChatGPT plan and, in a managed workspace, on administrator controls.
Help: help.openai.com/en/articles/…
#AI #AIagents
2
1
306
DY retweeted
A request that says GPTBot is not automatically GPTBot.
Mintlify open-sourced agent-traffic to classify HTTP requests as agent, ai-crawler, crawler, bot, human, or unknown. Official docs: it uses a curated dataset of user agents, headers, and operators' published verification sources, not a substring match alone. A coding agent fetching docs for a person, a training crawler, and a script mirroring llms.txt are different categories.
I have not wired this into production traffic. Official first checks:
npm i the official scoped package mintlify/agent-traffic (exact name is in the install doc)
From the docs, a user agent only check:
import { classify } from the official scoped package mintlify/agent-traffic
classify({ userAgent: 'Claude-User (claude-code/2.1.308; +support.anthropic.com/)' })
// category: 'agent', kind: 'coding', name: 'Claude Code', operator: 'Anthropic'
To catch impersonators, pass headers plus the real client IP and load published IP ranges:
import { classify, loadIpRanges } from the official scoped package mintlify/agent-traffic
const ipRanges = await loadIpRanges()
const result = classify(
{ headers: request.headers, ip: request.headers.get('cf-connecting-ip') },
{ ipRanges }
)
Official limits worth reading first: agents running inside a person's own browser are invisible at the HTTP level. Some products share an identity; docs say Codex has been reported to fetch as ChatGPT-User, and Cursor through Exa as ExaUser. Results name only what can be proven. Logs that keep only the user agent cannot catch agents that borrow a browser UA.
On Cloudflare Workers, load IP ranges inside the request handler, not at module scope.
Skill install if you want your coding agent to wire it: npx skills add mintlify/agent-traffic
Docs: agent-traffic.mintlify.app/i…
Categories: agent-traffic.mintlify.app/c…
Repo: github.com/mintlify/agent-tr…
#AI #AIagents
Stopping an Expo Simulators session stops the clock. It does not leave you a device you can reopen.
Official docs: Stopped and Errored are final states. Start a new session if you need the device again. A new remote device starts blank unless you give an application source. I have not run a cloud simulator session, so this is the official after-stop card, not a personal timing result.
What official docs say remains, and what you have to set up again:
Still available after stop:
- The session record: ID, name, tags, status, timestamps, activity events, and artifacts, until you delete the session
- Artifact file sizes, which appear only after the session stops
- Job logs and recordings you can still fetch with the saved session ID, for example:
eas simulator:get --id <session-id> --json --non-interactive
Cleared or gone after a confirmed stop:
- EAS CLI resets .env.eas-simulator so only the comment header remains. The local session ID is no longer in that file, which is why you should keep the ID if you still need artifacts
- REST API remoteConfig is null. The remote device and its controller endpoints no longer exist
- The next session will not have the previous session's in-memory agent context. You have to provide the app source and the prompts again
Before you start the next one, official start flags still matter: --max-duration-minutes (up to 40 minutes on Free, 115 on paid plans) and optional --max-idle-time-minutes, because a new session also keeps counting until you stop it or it hits a limit.
Add .env.eas-simulator to .gitignore. Do not print controller tokens in logs.
Docs: docs.expo.dev/eas/simulators…
CLI: docs.expo.dev/eas/simulators…
Security: docs.expo.dev/eas/simulators…
#AI #AIagents
1
1
298








