Friday roundup: 30 Oct 2026

Claude Code 2.1.314 adds a setting that blocks writes outside your working directories, the Claude Platform gets a beta usage and cost analytics API and a 128 MB request limit, and Codex CLI 0.166.0 adds GPT-6.1 Terra to its Vertex AI and Amazon Bedrock catalogs.

  1. Anthropic

    Claude Code 2.1.314

    Released 30 Oct 2026, 01:06 UK

    A new permissions.blockWritesOutsideWorkingDirectories setting makes Write, Edit and NotebookEdit refuse any path outside your working directories, in every permission mode.

    Gateway upstreams for Claude apps get an optional groups list for routing by IdP group, Bedrock gateway upstreams get guardrail_trace, and VS Code and the Desktop Code tab now compact idle conversations over 200K tokens.

    Fixes cover an "Invalid effort level" loop when an admin caps effort (it now lowers effort and retries), claude -p exiting 0 with no output after a failed startup (it now exits 1), gateways that reject tool_reference or beta headers (now retried without them), dotted model IDs like claude-opus-4.8 being misread behind gateways, and sandbox protections not applying to worktrees under -p --resume. Version 2.1.313 is folded into this release.

    Who it affects Anyone running Claude Code unattended or in CI. The write setting is worth turning on there, and scripts that call `claude -p` should check they now handle the non-zero exit when startup fails.

    Source: Claude Code 2.1.314 release ↗

  2. Anthropic

    Claude Platform Usage and Cost Analytics API, beta

    Released 29 Oct 2026

    New /v1/organizations/analytics/usage and /cost endpoints, behind the beta header analytics-usage-cost-2026-08-24, are available to Console and Enterprise organisations. They support filtering and grouping by model, workspace, API key, and user or service account.

    The older Usage and Cost API and the Enterprise usage and cost report endpoints are now legacy but still supported.

    Who it affects Teams that charge AI costs back to projects or need to spot a runaway agent by key or user. Anyone on the older endpoints should note they are now legacy.

    Source: Claude release notes ↗

  3. Anthropic

    128 MB request size for the Messages and Token Counting APIs

    Released 28 Oct 2026

    The maximum request size for the Messages and Token Counting APIs on the first-party API went from 32 MB to 128 MB.

    Who it affects Developers sending large documents or batches of images in a single call through the first-party API.

    Source: Claude release notes ↗

  4. OpenAI

    Codex CLI 0.166.0

    Released 29 Oct 2026, 19:47 UK

    This release adds keyboard-driven transcript selection with word, line and page motions, search and configurable shortcuts, and consistent Markdown in async questions and answers.

    GPT-6.1 Terra is added to the bundled, Vertex AI and Amazon Bedrock model catalogs. There is also codex invoke --resume for unloaded threads, clearer explanations when a request is blocked for safety reasons with safer follow-up prompts, and codex doctor --network-only for config, connectivity and WebSocket checks.

    Fixes keep approval and permission settings and reasoning-token accounting across resume and fork, so threads don't compact early, and TUI reconnection is more robust.

    Who it affects Teams that route models through Vertex AI or Amazon Bedrock, who can now select GPT-6.1 Terra inside Codex, and anyone resuming or forking long-running threads.

    Source: Codex CLI 0.166.0 release ↗

← All news