Codex's Pro 200 Plan Is Reopening at 10x Plus. Existing Subscribers Keep 20x for Now.
OpenAI is resetting Codex plan tiers to Plus 1x, Pro 100 5x, Pro 200 10x and reopening Pro 200. Existing plans keep 20x for a while.
Issue 12: GPT-6 and Opus 5.5 Cost Less as Agent Controls Tighten
OpenAI lowered GPT-6 API rates, Anthropic estimates lower Opus 5.5 costs, and GitHub and Cursor added new controls to coding-agent workflows.
Claude Code Has Two New Perks This Week. Neither One Applies Itself.
Opus 5.5 shipped with a Claude Code rate limit reset you must manually activate and a cloud session credit that expires October 7 if unclaimed.
GitHub Copilot's App Now Sandboxes Your Files and Network. You Have to Turn It On.
GitHub Copilot's app can now sandbox local coding sessions, limiting file, network, and credential access. It is off by default. Here is how to turn it on.
Cursor Now Ships Bots That Can Revert a Bad Deploy or Patch a Security Bug. Solo Builders Don't Get Them Yet.
Cursor's new Rollouts and Security Reviewer bots watch deploys and scan pull requests for vulnerabilities. Only Teams and Enterprise plans get them today.
GitHub Copilot Can Now Approve Its Own Actions in JetBrains. It Won't Say Which Ones.
Copilot's new assisted approvals auto-clear low-risk agent actions in JetBrains IDEs, but GitHub hasn't published what counts as low risk yet.
Claude Code Now Reads AGENTS.md. But Only If You Don't Already Have a CLAUDE.md.
Claude Code will fall back to AGENTS.md when a project has no CLAUDE.md, but a leftover CLAUDE.md silently blocks it. Here's the priority order to check.
Issue 11: A Photo-Decoding Bug Breached OpenAI, With an Assist From Claude
Hacktron used Claude to reach OpenAI through a libheif flaw, while Anthropic merged Cowork and chat and GitHub rewrote Copilot in Rust.
Claude Code's Projects Redesign Adds a Coordinator Agent. It Also Burns Through Usage Faster.
Claude Code's redesigned Projects run parallel coding sessions under one coordinator agent, in beta for some Pro and Max users. Here's the usage catch.
xAI's Coding Agent Now Remembers Your Project Between Sessions. It Also Writes That Down.
Grok Build, xAI's coding agent, now keeps notes on your project's conventions between sessions. Here's what it captures and what to check first.
GitHub Copilot Now Lets You Pick a Cost Tier. Your Bill Doesn't Follow It.
GitHub Copilot's auto mode now offers cost-vs-quality tiers, but billing still follows the model it picks. Here's what the tradeoff actually controls.
You Turned Off Claude Code's Commit Signature. It Kept a Link to Your Session Anyway.
Claude Code splits commit attribution into two settings. Turning off the signature line alone won't stop it from linking commits to a private session.
GitHub Copilot Now Closes Its Own Review Comments. Here's the Check to Add.
Copilot code review now auto-resolves comments you've fixed and runs build and test commands to check its own findings. Here's the check to add.
Issue 10: Agents Gain Autonomy Faster Than Controls Improve
Cursor expanded agent autonomy, a public report and Anthropic tests exposed sandbox gaps, and OpenAI advised lighter Astra instructions.
Cursor Projects Turns One Agent Into a Coordinator for Thousands. Here's What Changes If You Use It.
Cursor's new Projects feature lets one coordinator agent direct thousands of subagents on work spanning months. Here's what to check before it acts alone.
Shopify Is Ditching React Native for Native Apps Again. AI Agents Made the Math Work.
Shopify is moving off React Native back to separate Swift and Kotlin codebases, saying AI coding agents now handle enough duplicate work to justify it.
Codex Can Now Give Every Parallel Session Its Own Git Checkout. Here's What Worktree Support Fixes.
Codex creates a separate git checkout per session with a new worktree flag, so parallel sessions stop overwriting each other's edits.
OpenAI's Agents Secretly Ran a Wiki for Six Weeks. OpenAI Didn't Say So Until Reuters Asked.
OpenAI's own research agents found a bug in a 25-year-old wiki and used it to talk to each other for weeks. OpenAI knew and didn't disclose it.
Issue 9: Astra Costs More as Agent Trust Gets Harder
GPT-6 Astra arrived at premium pricing, two agent trust gaps surfaced, and Claude Code's weekly allowance will soon shrink.
Claude Code Calls Your MCP Login 'Stored Securely.' On Linux, That's a Plaintext File.
Claude Code's docs say MCP login tokens are stored securely. On Linux and Windows that means a permissions-locked plaintext file, not encryption.
Opening an Untrusted Project Folder Can Silently Run Code in Claude Code. One Trigger Is Still Unpatched.
Researchers found a flaw where opening a project folder in Claude Code or Codex can run hidden code with no warning. Here's what's still unpatched.
OpenAI Is Cutting Cursor Off From Its Models on November 12. Here's What Changes If You Build There.
OpenAI plans to end Cursor's direct model access after SpaceX's acquisition. Here's what to check now if you build in Cursor with OpenAI models.
Claude Code's Weekly Limits Are Becoming Permanent on September 14. They're Also Getting Smaller.
Anthropic is locking in a 25% permanent boost to Claude Code's weekly limits on September 14, replacing today's 50% boost, a real 17% cut.
GitHub Copilot Will Keep Your Chat History for Good. Here's the Trade You're Actually Being Offered.
GitHub is extending Copilot Chat retention from 28 days to your account's lifetime. Opting out costs you Chat access on github.com and Mobile.