An Expired Login Was Nudging Max Users Toward Paid Usage — How Claude Code v2.1.227 Closed a Billing Mix-Up and a Total CI Wipeout

Claude Code v2.1.227 is mostly bug fixes. But it closes two holes that are easy to overlook in unattended operation: one that wrongly nudged Max users toward pay-as-you-go billing via an expired token, and one that wiped out every Bash command in CI on GitHub runners. Here's what to double-check.

Share
An Expired Login Was Nudging Max Users Toward Paid Usage — How Claude Code v2.1.227 Closed a Billing Mix-Up and a Total CI Wipeout

The latest release of Claude Code, v2.1.227, is out. There are no headline features; the substance is mostly bug fixes. But two of them bear directly on unattended agents and on how billing is handled. Both are the kind of flaw that breaks quietly and is easy to miss, and left alone they can affect your bill or your CI.

An Expired Login Was Nudging Max Users Toward Paid Usage

This is the fix with the greatest real-world impact this time around. When a session began with the login token in an expired state, feature flags were being evaluated without taking the user's subscription plan into account.

As a result, Max plan users could be wrongly prompted to enable pay-as-you-go usage credits for the higher-end Fable model. In effect, a doorway to extra charges was being offered to people who shouldn't have needed to pay more at all.

Access to Fable has gradually shifted from being bundled for free to a metered usage-credit system, so misjudging the plan leads straight to billing confusion. v2.1.227 fixes this so that flags are evaluated with the subscription plan in mind even when the token has expired.

"Every Command Fails" in CI — the claude-code-action Pitfall

The other fix matters for automation. Using claude-code-action with allowed_non_write_users on a GitHub-hosted runner caused every single Bash command to fail.

Since it was a total failure rather than a partial one, the whole pipeline grinds to a halt. If you're running unattended, this is the kind of outage that eats up your time in narrowing down the cause. This release makes commands run successfully even in that configuration.

Smaller Fixes That Catch the Stragglers

  • Conversation rewind: Fixed a bug where /tui would bring back a conversation that had been rewound to a point before the first message.
  • Slash command menu: Only the selected row is now highlighted in blue, and matched characters are shown in bold rather than a different color. Emoji and accented names now keep their glyph shapes.
  • Perceived speed: Reduced event-loop stalls that occurred when displaying candidates for files that weren't found and when checking the size of @-mentions, easing the hitches.

What to Check

The update itself arrives through the normal upgrade process. On top of that, it's worth confirming just two things to be safe.

  • Billing: If you recall being prompted on Max or another higher-end plan to enable Fable credits, review your settings to make sure you didn't enable them unintentionally.
  • CI configuration: If you run claude-code-action with allowed_non_write_users, upgrade to v2.1.227 and confirm that your Bash steps pass.

Behind all this convenience, the agent's own state judgments increasingly sway billing and CI. Bugs like this time's — "losing sight of the plan" or "every command failing" — would be caught quickly if a person were watching constantly, but under unattended operation they can just as easily slip by. It suggests that the more you push automation, the more you need a practice of following the changelog so fixes like these don't get missed.

References: Claude Code changelog (official) / anthropics/claude-code CHANGELOG.md / Releasebot: Claude Code Updates

Read more

Making It Wait for "Jobs That Run Over an Hour": Codex 0.152 Adds Ceiling Dials for MCP Output Volume and Execution Time, and Turns the Planning Tool Off by Default

Making It Wait for "Jobs That Run Over an Hour": Codex 0.152 Adds Ceiling Dials for MCP Output Volume and Execution Time, and Turns the Planning Tool Off by Default

Codex v0.152.0 on August 31 and its next-day fix release added explicit ceilings on MCP tool output volume and execution time, and switched the planning tool off by default. Here's a rundown of the changes that matter for long-running unattended and semi-autonomous agent operation.

By FF
The CLI's Default Model Just Swapped In a Million-Token Brain — Claude Code v2.1.257 Makes Fable 5.1 the Standard and Adds a 'Containment Escape' Checkpoint to Auto Mode

The CLI's Default Model Just Swapped In a Million-Token Brain — Claude Code v2.1.257 Makes Fable 5.1 the Standard and Adds a 'Containment Escape' Checkpoint to Auto Mode

Claude Code v2.1.257, released September 1, 2026, swaps its default model to Fable 5.1 with its one-million-token context. It also adds guardrails to auto mode that stop credential retrieval and out-of-scope reads from slipping through. Here's a rundown of the changes that matter to developers.

By FF
"This Is an Authorized Exercise"—How the Aurora Ransomware Gang Insisted, While Making Cursor's AI Agent Do the Actual Intrusion Work

"This Is an Authorized Exercise"—How the Aurora Ransomware Gang Insisted, While Making Cursor's AI Agent Do the Actual Intrusion Work

Gambit Security and CloudSEK report that the ransomware group Aurora abused Cursor's AI agent for real intrusion work. Posing the tasks as an "authorized exercise" to slip past the safeguards, they had it handle reconnaissance and privilege takeover on the back of stolen credentials—a warning that a

By FF
One in Three Companies Now Choose to Build Rather Than Buy — McKinsey Measures How Coding Agents Are Reshaping the Procurement Decision

One in Three Companies Now Choose to Build Rather Than Buy — McKinsey Measures How Coding Agents Are Reshaping the Procurement Decision

McKinsey's annual survey found that about 30% of respondents passed on buying software because they could build it in-house with coding agents. We unpack the procurement shift from buying to building — and the current reality that productivity is up while profits stay flat.

By FF