What's New on AI Developer Toolkit

The AI Developer Toolkit changelog records the Claude Code, Codex and Cursor releases that change how an agent setup behaves (default models, removed flags, permission defaults), the fix for each, and every guide this site published. As of 12 October 2026, the changes needing action are the new Claude Sonnet 5.5, Haiku 5.5 and GPT-6.1 Sol, plus the retirements of GPT-5.5 in Codex (2026-10-14) and Claude Sonnet 4.5 (2026-11-30).

A guide written in July tells you to run codex exec --full-auto, and your CI job now fails with an unexpected-argument error. A teammate’s Claude Code session runs Opus 5 while yours runs Opus 5.5, and neither of you changed a setting. This page is for the developer or tech lead who owns a team’s shared agent configuration (CLAUDE.md, AGENTS.md, .claude/settings.json, Codex’s config.toml, the CI jobs that call an agent) and needs to know which release caused the change and what to edit.

What you get from this changelog as the owner of a team’s agent config

• Tables of the October and September 2026 changes that break or silently change a setup, each with the version that introduced it and the fix.
• A ten-minute check that tells you whether your machines and CI are behind this page.
• Three copy-paste prompts: one audits a repository for stale model IDs and removed flags, one applies the fixes on a branch with evidence, and one filters an upstream changelog down to what touches your config.
• A way to prove an upgrade did not break your agent workflow without reading every transcript.
• A dated record of what this guide shipped, so you can tell whether a page predates the release it describes.

Which changes from October 2026 need action?

Every row was checked on 2026-10-12 against the vendor’s changelog, documentation, release tags or the CLI binaries (claude 2.1.296, codex-cli 0.162.1; Cursor 3.24 from Cursor’s release notes), and covers 2026-09-27 to 2026-10-12. Prices and context windows live on one page, the AI model comparison for coding agents, so this table names models without repeating their prices. The September table further down stays as the record of 2026-09-26; where October changed a September row, the October row says so.

Claude Code

Change	Since	What to do
The stable channel catches up to v2.1.287	stable 2.1.287 and latest 2.1.296 on 2026-10-12	AGENTS.md reading (v2.1.277), the Opus 5.5 default (v2.1.280), auto mode as the starting mode (v2.1.283) and Sonnet 5.5 (v2.1.284) now apply on both channels, so a team on stable that missed the September changes gets them now. On 2026-10-12, changes from v2.1.288 on were on latest only; on 2026-10-17 stable is v2.1.292, so changes from v2.1.293 on, such as the haiku alias, are the ones on latest only. Compare claude --version across machines.
Claude Sonnet 5.5 is the sonnet alias on the Anthropic API	v2.1.284 (2026-09-28), both channels; Sonnet 5.5 needs v2.1.284 or later	Remove hard-coded claude-sonnet-5 pins unless your evals need them; Sonnet 5 is now legacy. In Claude Code, Opus 5.5, Sonnet 5.5 and Haiku 5.5 all start at medium effort (Opus 4.7 at xhigh, other models that support effort at high). sonnet still resolves to Sonnet 4.6 on Claude Platform on AWS and to Sonnet 4.5 on Bedrock, Google Cloud and Foundry. Auto mode’s classifier still runs on Claude Sonnet 5 by default, not on your /model choice. If your own API code runs Sonnet with thinking off, move to thinking: {"type": "between_tools"} first: disabled returns a 400 on Sonnet 5.5.
Claude Haiku 5.5 is the haiku alias on the Anthropic API	v2.1.293 (2026-10-07), latest channel only	Haiku 5.5 needs v2.1.293 or later, so on stable (2.1.292 on 2026-10-17) haiku still means Haiku 4.5; Anthropic’s setup page says to move to latest to run a newly launched model. Other providers keep Haiku 4.5, now legacy. Haiku 5.5 has an effort setting (medium by default) and a 1M window on every plan with no [1m] suffix. Auto mode now supports Haiku 5.5, but not Haiku 4.5 or Sonnet 4.5 on any provider. Its price depends on prompt length, so re-price subagents that run on haiku from the models hub.
Claude Sonnet 4.5 is deprecated and retires on 2026-11-30	Deprecated 2026-09-30; retires 2026-11-30 on Anthropic-operated platforms	Anthropic’s deprecation page applies that date to the Claude API, Claude Platform on AWS and Microsoft Foundry; Bedrock and Google Cloud set their own dates. Claude Code’s model-config page still names Sonnet 4.5 as Foundry’s default model and as the sonnet alias on Bedrock, Google Cloud and Foundry (checked 2026-10-12). On those providers, pin a model explicitly before the date; Anthropic’s recommended replacement is claude-sonnet-5-5.
Ultracode is its own toggle	v2.1.284 (2026-09-28), both channels	Turning ultracode on or off with /effort or the ultracode setting leaves the effort level unchanged; --effort ultracode still sets xhigh. An effort cap no longer makes it unavailable: it stays on at the capped level, and Claude Code’s model-config page lists only workflows turned off or a model without xhigh as reasons it is unavailable. Review any policy that used a cap to keep ultracode off. See dynamic workflows and ultracode.
claude -p and Agent SDK sessions start in auto mode on third-party providers	v2.1.285 (2026-09-29), both channels	With no permission mode configured, claude -p and Agent SDK sessions on third-party providers, or with telemetry off, start in auto mode. Elsewhere they still start in Manual, as do sessions in an organization with the HIPAA configuration (from v2.1.285) and any session whose settings file sets disableAutoMode: "disable". Pass --permission-mode manual in CI jobs that must not approve actions on their own.
Unattended runs are stricter	--bare in v2.1.286 (2026-09-30); background-command limit in v2.1.285 (2026-09-29), both channels	--bare connects only the MCP servers named on the command line, sends no system reminders and starts no background tasks. Background shell commands stop after their timeout (default 30 minutes, maximum 2 hours); from v2.1.288 (both channels on 2026-10-17) that limit applies only in unattended sessions (-p, the Agent SDK, CI and cloud). Check CI jobs that use --bare or leave a dev server running in the background.
claude project purge is now claude purge	v2.1.288 (2026-10-02), both channels on 2026-10-17	The old name still works and prints a notice. claude purge deletes all Claude Code state for a project. Switch scripts once every machine runs v2.1.288 or later; stable is 2.1.292 on 2026-10-17.
The earlier Claude in Slack app is retired	2026-10-05	In workspaces paired with a Claude organization, the earlier app (shown as Legacy in admin settings) no longer answers in channels; direct messages are not affected. Claude Tag (Public Beta; Team and Enterprise plans, Anthropic’s first-party service only) replaces it. Pro and Max accounts keep the earlier app only in workspaces no organization has connected to Claude Tag. See Claude Tag.
Max and Team plans include monthly API credits	2026-10-07	Max 5x gets $100, Max 20x $200, and Team $20 per Standard seat or $100 per Premium seat, pooled and capped at $500. The credits cover the Claude API (Messages and Batches), Claude Managed Agents and the Claude Agent SDK. They do not cover interactive Claude Code in the terminal, IDE, desktop or web, and they do not change your usage limits. Claim them by linking one Claude Console organization; they do not roll over. Free, Pro and Enterprise plans are not eligible.

Codex

Change	Since	What to do
GPT-6.1 Sol replaces GPT-6 Astra as the bundled default	CLI 0.159.1 (2026-09-29); restated in 0.161.0 (2026-10-07)	CLIs before 0.159.1 still bundle GPT-6 Astra as the default; it held that role from 0.153.4 and is second in the picker now. Sol’s default effort in Codex is low; raise effort before you switch models. A model pin in config.toml overrides the default, so a pin to gpt-6-astra keeps Astra. Since 0.162.0, new TUI threads respect the server’s model default unless you pass an explicit launch override. OpenAI’s Codex docs recommend Sol for complex coding and agentic work and Astra when a task needs the strongest capability across steps and tools.
GPT-5.5 leaves ChatGPT, ChatGPT Work and Codex	2026-10-14, all plans (OpenAI’s models page, checked 2026-10-12)	Change any model = "gpt-5.5" in config.toml or CI before that date, so the file says what runs. The OpenAI API is not affected. The bundled catalog in 0.162.1 carries no retirement date for GPT-5.5 (retirement_at is null), so the date comes from OpenAI’s page, not from the binary. Codex’s own upgrade prompt for GPT-5.5 points to GPT-6 Sol.
Codex in Slack is @ChatGPT	OpenAI’s Slack page, checked 2026-10-12	The page is now titled “Use ChatGPT in Slack”: mention @ChatGPT in a channel to start a request. With Cloud delegation enabled, ChatGPT can send repository work to a separate Codex Cloud task and return the result to the thread; that needs a published environment shared with the workspace. The @Codex sentence the page carried on 2026-08-28 is gone. Linear still uses @Codex and GitHub review still uses @codex review. Update runbooks that tell people to mention @Codex in Slack. See Slack and Linear.
Codex Cloud environments are published and reusable	DevDay (2026-09-29); OpenAI’s cloud pages, checked 2026-10-12	Create or select an environment, let Codex prepare your project, then review and publish it; new tasks start in isolated workspaces from the prepared filesystem. The earlier setup is now Codex Cloud (Legacy), which “continues to support environments for Code Review and the Linear and GitHub integrations”. Check which kind of environment each of your integrations uses. See cloud environments.
The bundled plugin-creator skill and tui.prompt_suggestions are removed	CLI 0.159.0 (2026-09-29)	Remove prompts, docs and config.toml entries that call the plugin-creator skill or set tui.prompt_suggestions. The bundled sample skills in 0.162.1 are imagegen, openai-docs, review-agent, skill-creator and skill-installer.
Terminal input needs approval by default	CLI 0.158.0 (2026-09-28)	Terminal input approval is on by default for commands running with elevated permissions. Expect approval prompts for those commands, and test your attended and unattended flows after you update.
September’s removed flags are still removed	Checked on 0.162.1 (2026-10-12)	codex --help is identical to 0.157.1: no top-level flag or subcommand was added or removed. --full-auto is still an unexpected argument, -a still accepts only on-request and never, codex exec still has no -a, and mcp_2026_07_28 is still under development and off in codex features list. There is no /loop command in Codex 0.162.1. The September fixes still apply.

Cursor

cursor.com was reachable on 2026-10-12, so these rows come from Cursor’s documentation, help pages and release notes. They take precedence over the September note below that Cursor could not be re-verified.

Change	Since	What to do
Version numbers live in Cursor’s release notes	Cursor 3.24 (2026-10-09)	The marketing changelog stopped printing version labels after 3.11 (2026-07-10); the release notes number every minor version of Cursor IDE and the Agents Window, and the CLI uses date versions (v2026.10.08). Read your version in the About dialog and compare it with Cursor’s release notes. 3.24 opens the Goal, queue, debug, canvas and browser panels from pills above the prompt, makes Auto-review stop risky deletes for your approval, and adds the managed policy AgentRemoteWorkspaceAuthorities, which lets admins restrict Agent to approved remote workspaces.
Run modes have names: Auto-review, Allowlist and Run Everything	Auto-review has been the recommended default since 3.6 (2026-05-29)	Find them under Settings > Agents > Approvals & Execution. Auto-review runs allowlisted calls at once, runs other shell commands in the sandbox when possible and sends the rest to a classifier model. Cursor’s docs name that model as Gemini 3.5 Flash Lite with Claude 4.5 Haiku as the fallback, and say blocking Claude 4.5 Haiku “can disable Auto-review”. Cursor also says Auto-review is not a security boundary. Since 3.5, Ask Every Time is deprecated (use Allowlist with an empty allowlist) and Run in Sandbox is folded into Allowlist. Cloud Agents do not use run modes. See essential Cursor configuration.
AGENTS.md and CLAUDE.md are read; .cursorrules is legacy	Cursor’s rules docs and help pages, checked 2026-10-12	Cursor supports AGENTS.md in the project root and in subdirectories, with the more specific file taking precedence, and reads CLAUDE.md the same way, always applied. Project rules are .mdc files in .cursor/rules; a plain .md file there is ignored. Cursor’s help says the .cursorrules file “is legacy and will be deprecated”: migrate it to a project rule set to Always Apply. Rules apply to Agent chat, not to Tab completion, Inline Edit or Bugbot reviews. See project rules.
Claude Opus 5.5, Sonnet 5.5 and Haiku 5.5 are in Cursor; no GPT-6 model is	Opus 5.5 in the week of 2026-09-21, Sonnet 5.5 in the week of 2026-09-28	Cursor’s release notes list Opus 5.5 and Sonnet 5.5 as new models, and cursor.com has a model page and a price for Haiku 5.5. Version 3.24 adds Opus 5.5 and Sonnet 5.5 on AWS Bedrock for bring-your-own-key setups. Cursor’s newest OpenAI models are GPT-5.6 Sol, Terra and Luna, and its models page lists no GPT-6 model (neither GPT-6.1 Sol nor Astra; checked 2026-10-12). Its own model pool is Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5, and Cursor calls Grok 4.7 its flagship. Prices are on Cursor’s models and pricing page.
The CLI command is agent	Cursor’s CLI docs, checked 2026-10-12	Cursor’s docs name the command agent: check it with agent --version, update with agent update, and run headless with agent -p. The name cursor-agent appears on no page of Cursor’s docs, help or CLI changelog, so check the scripts and CI jobs that call it.
Remote control for local agents	2026-10-06	You can see and reply to the agents running on your computer from the Cursor iOS app. It is on by default except for Enterprise organizations, where admins enable it at Org settings > Security and identity > Remote control. Cursor says remote control “doesn’t move your agents anywhere” and doesn’t require cloud agents. Check that the default fits your security policy.

Cursor’s /goal is still labelled “rolling out” in its agent overview, and /loop is a built-in skill (checked 2026-10-12).

Three changes cut across tools:

• MCP specification 2026-07-28. Claude Code now also negotiates the new version by default with local (stdio) servers on every install, including Bedrock, Google Cloud and Foundry, from v2.1.292 on both release channels; set MCP_PROTOCOL_NEGOTIATION=legacy to opt out. Codex 0.162.1 still does not negotiate it (mcp_2026_07_28 is under development and off, checked 2026-10-12). MCP Server Cards (SEP-2127), a format that describes a server before a client connects, became Final on 2026-10-06; the specification version is unchanged. Test your MCP server against both protocol versions. See MCP best practices.
• GitHub Copilot’s four retirements happened on 2026-10-02. GitHub removed Kimi K2.7 Code, Gemini 3.5 Flash, Gemini 3.6 Flash and Claude Opus 4.7 from its model tables. GitHub’s announcement suggests Claude Opus 5.5, not Claude Opus 5 as the cross-tool note in the September section below says, as the replacement for Opus 4.7. Copilot added Claude Sonnet 5.5 (2026-09-28), GPT-6.1 Sol (2026-09-29, not on the Pro plan) and Claude Haiku 5.5 (2026-10-07), and GitHub’s docs list no further retirement (checked 2026-10-12). See the GitHub Copilot comparison.
• The official Terminal-Bench 4.0 leaderboard. On the official board (read 2026-10-12), Claude Opus 5.5 at max effort in Claude Code scores 64.8% ± 3.1 and Claude Sonnet 5.5 61.8% ± 2.9; GPT-6 Astra and GPT-6.1 Sol, both at max effort in Codex, tie at 58.2% (± 2.8 and ± 3.1). The Opus 5.5, Sonnet 5.5 and GPT-6.1 Sol rows were created on 2026-10-06. Correction: GPT-6 Astra has had rows on this board since 2026-09-03; the 2026-09-26 read of the GitHub submissions folder missed them. Vendor launch tables report different figures for the same models, so keep the two sources apart. See the benchmarks page.

Which changes from September 2026 need action?

Every row was checked on 2026-09-26 against the vendor’s changelog, release tags or the installed CLIs (claude 2.1.283, codex-cli 0.157.1). Prices and context windows live on one page, the AI model comparison for coding agents, so this table names models without repeating their prices. These tables are the record of 2026-09-26; the October section above says where a later release changed a row.

Claude Code

Change	Since	What to do
Claude Opus 5.5 is the default model on every paid plan and the API	v2.1.280, latest channel (2026-09-22)	Remove hard-coded claude-opus-5 pins unless your evals need them. Opus 5.5 runs at medium effort by default: raise it with /effort before you switch models.
The stable channel lags behind	stable was v2.1.274 on 2026-09-26	On 2026-09-26 stable was v2.1.274: there Pro and Team Standard still defaulted to Sonnet 5, other plans to Opus 5, and Opus 5.5 was unavailable. Check where each channel is now with npm view @anthropic-ai/claude-code dist-tags. Put the whole team on one channel with autoUpdatesChannel.
Claude Fable 5.1 replaces Fable 5	v2.1.257 (2026-09-01)	The fable alias now resolves to Fable 5.1. Fable is never a default; select it deliberately with /model fable, and expect it to bill to usage credits on some plans.
Auto mode is the starting permission mode	v2.1.283, latest channel (2026-09-25)	Interactive terminal and VS Code sessions start in auto mode. claude -p and the Claude Agent SDK still start in Manual. To keep Manual organisation-wide, set permissions.disableAutoMode: "disable" in managed settings. Sessions on a model auto mode does not support (for example Haiku, or Sonnet 4.5, the default on Microsoft Foundry) also start in Manual.
AGENTS.md is read when a project has no CLAUDE.md	v2.1.277 (2026-09-18); v2.1.281 on Bedrock, Google Cloud, Foundry and gateways	Read the AGENTS.md in repositories without a CLAUDE.md: Claude now loads it. Toggle this under Project instructions in /config.

Codex

Change	Since	What to do
GPT-6 Astra is the bundled default	CLI 0.153.4 (2026-09-04); needs 0.153.0+	Update CLIs older than 0.153.0. Astra’s Codex default effort is low; raise effort before switching models.
GPT-6 Sol and GPT-6 Luna join the picker	CLI 0.156.1 (2026-09-23); need 0.155.0+	Use Sol for everyday work and Luna for cheap, high-volume tasks. There is no GPT-6 Terra: Terra exists only as gpt-5.6-terra.
GPT-5.4 is retired in Codex	2026-08-31	A config that pins gpt-5.4 is migrated to GPT-6 Sol. Change model in config.toml so the file says what actually runs.
codex exec --full-auto is removed	CLI 0.147.0 (2026-08-07)	Use a permission profile (-c default_permissions=":workspace", beta), --approve-for-me, or --sandbox workspace-write. Do not combine a profile with --sandbox: OpenAI says the two systems do not compose.
-a accepts only on-request and never	checked on 0.157.1	Replace -a on-failure and -a untrusted. Put the flag before the subcommand: codex -a never exec …, because codex exec has no -a of its own.
codex mcp-server is removed	CLI 0.154.0 (2026-09-09)	Codex can no longer run as an MCP server. Use codex exec, the Codex SDK, or codex app-server (experimental). Claude Code still offers claude mcp serve.

Cursor

Cursor’s changelog and documentation could not be re-verified for September 2026: cursor.com was unreachable from the environment this page was checked in on 2026-09-26. The last verified Cursor facts date from 2026-08-28 and are summarised on the Cursor changelog page. Read Cursor’s own changelog before you act on any Cursor version, model or price claim, including ones on this site.

Two changes cut across tools:

• MCP specification 2026-07-28 removed the initialize handshake and the Mcp-Session-Id header. Claude Code negotiates the new version by default with direct HTTP servers (all installs by v2.1.274); Codex 0.157.1 does not yet (mcp_2026_07_28 is off in codex features list). If you maintain an MCP server, test it against both protocol versions. See MCP best practices.
• GitHub Copilot retires four models on 2026-10-02: Kimi K2.7 Code (replaced by Kimi K3), Gemini 3.5 Flash and Gemini 3.6 Flash (replaced by Gemini 3.8 Flash), and Claude Opus 4.7 (replaced by Claude Opus 5). Update any Copilot model policy that names them.

Check whether your setup is behind this page

Run these on a developer machine and in one CI job. The whole check takes about ten minutes and changes nothing.

1. Record the versions you actually run, in a terminal:

   claude --version
   codex --version

   For Cursor, read the version in its About dialog and compare it with Cursor’s release notes. On 2026-10-12 the current versions are Claude Code 2.1.296 (latest) and 2.1.287 (stable), Codex 0.162.1 and Cursor 3.24; on 2026-10-17 Claude Code is 2.1.302 (latest) and 2.1.292 (stable).

2. Find the Claude Code release channel. Look for autoUpdatesChannel in ~/.claude/settings.json and in managed settings; "stable" means you are about a week behind latest. On 2026-10-17 stable is 2.1.292 and latest is 2.1.302. To choose a channel, set autoUpdatesChannel to "stable" or "latest" in settings; claude install stable installs the stable build now.

3. List the models your Codex binary knows and the state of its feature flags:

   codex debug models --bundled
   codex features list

4. Run both health checks. They read your settings files and report problems before an agent run does:

   claude doctor
   codex doctor

5. Search the repository for the removed flags and legacy model pins from the tables, from the repository root:

   rg -n -e '--full-auto' -e '-a (on-failure|untrusted)' -e 'codex mcp-server' \
     -e 'gpt-5\.[45]' -e 'claude-sonnet-4-5' -e 'claude project purge' \
     -e 'claude-(opus|fable)-5([^-]|$)' .

   The last pattern matches claude-opus-5 and claude-fable-5 but not claude-opus-5-5. The pattern gpt-5\.[45] matches both gpt-5.4 and gpt-5.5; the retirement on 2026-10-14 does not affect the OpenAI API, so change only the pins that Codex reads.

6. Update on a branch, not on main: claude update and codex update, then repeat steps 1–4 and run the prompts below.

Copy-paste prompts for keeping agent config current

Run the first prompt in Claude Code or Codex from the repository root. It reads and reports; it edits nothing, so you can run it with read-only permissions.

Copy-paste prompt: audit the repository for stale agent config

Audit this repository's agent configuration for changes released up to 2026-10-12.
Read CLAUDE.md, AGENTS.md, .claude/settings.json, .mcp.json, any Codex config.toml,
.github/workflows/*.yml and scripts/. Report every occurrence of:
- `codex exec --full-auto` or `--full-auto` (removed in Codex CLI 0.147.0)
- `-a on-failure` or `-a untrusted` on the codex command line (only on-request and never are accepted in 0.157.1–0.162.1), and `-a` placed after `exec`
- `codex mcp-server` (removed in Codex CLI 0.154.0)
- the bundled `plugin-creator` skill or the `tui.prompt_suggestions` setting (removed in Codex CLI 0.159.0)
- a model pinned to gpt-5.4 (retired in Codex on 2026-08-31), gpt-5.5 (leaves Codex on 2026-10-14), claude-sonnet-4-5 (retires 2026-11-30 on Anthropic-operated platforms), claude-opus-5 or claude-fable-5 (now legacy)
- `claude project purge` (renamed `claude purge` in Claude Code v2.1.288; the old name still works)
- the word "workflow" used as the Claude Code dynamic-workflow trigger (replaced by "ultracode" in v2.1.160)
For each hit print file:line, the exact line, the replacement, and one sentence on the risk
of leaving it. Do not edit any file. End with a table sorted by risk, CI files first.

The second prompt applies the fixes and produces the evidence a reviewer needs, so the pull request can be approved on its checks.

Copy-paste prompt: apply the fixes on a branch with evidence

Create a branch named chore/agent-config-2026-10. Apply only the replacements from the
audit table you produced earlier in this session (if there is none, run the audit
first); change nothing else. After each file, run the checks this repo
uses in CI (read the workflow files to find them: tests, lint, type check). If a check
fails because of your change, fix it; if it fails for an unrelated reason, stop and tell
me. Finish with: the list of commands you ran with their exit codes, `git diff --stat`,
and a three-line pull request description that names each removed flag or model and
the release that removed it.

The third prompt turns a long upstream changelog into the handful of entries that touch your configuration. Replace 2.1.292 (the stable version on 2026-10-17) with the version your team runs; npm view @anthropic-ai/claude-code dist-tags shows the current channel versions.

Copy-paste prompt: filter an upstream changelog to what affects us

Our team runs Claude Code 2.1.292 (stable channel). Read the Claude Code changelog at
https://code.claude.com/docs/en/changelog.md from 2.1.293 up to the version `claude --version`
reports here. List only the entries that change a default (model, effort, permission mode),
remove or rename a command, flag or settings key, or change what gets loaded into context.
For each: the version, the change in one line, and whether .claude/settings.json, CLAUDE.md
or our CI workflows are affected, quoting the line if so. Skip bug fixes and UI polish.

How to prove an agent tool upgrade did not break your workflow

Nobody on the team should have to read transcripts to approve a version bump. Make the upgrade a change like any other, with checks that fail loudly:

1. Pin the versions CI installs. Install an exact version, for example npm install -g @anthropic-ai/claude-code@2.1.296 or npm install -g @openai/codex@0.162.1, so CI only changes when someone bumps the pin in a pull request.
2. Run one fixed smoke task headlessly on both versions. Pick a small task with a known correct outcome, such as adding a test for an existing function, and run it with claude -p … --output-format json or codex exec … --output-schema schema.json. The gate is that your test suite passes afterwards, not that the transcript looks right.
3. Re-run your evals when the default model changes. A new default model is a model switch you did not choose. Follow the 48-hour evaluation playbook for a new model and the evals for coding agents guide before you accept it for unattended work.
4. Let the config owner sign off on evidence. The tech lead who owns the shared configuration approves the bump when tests, lint, type checks and the smoke task are green on the new version. If something regresses, revert the pin; the previous version is one commit away.

What breaks after an agent tool update, and how to recover

Teammates see different default models. Cause: some machines follow Claude Code’s stable channel and others latest, or Homebrew’s claude-code cask (which tracks stable) sits next to a native install. On 2026-10-12 both channels default to Opus 5.5, but haiku resolves to Haiku 5.5 only from v2.1.293 (latest), and Microsoft Foundry still defaults to Sonnet 4.5. Recovery: set one autoUpdatesChannel for the team in managed settings, and pin a model with --model where a result must be reproducible.

A CI job fails with “unexpected argument” or “invalid value”. Cause: a removed Codex flag (--full-auto, -a on-failure) or -a placed after exec. Recovery: apply the Codex rows of the tables above, then pin the Codex version so the next removal arrives in a pull request instead of in a failed deploy.

Codex reports that GPT-5.4 is no longer available. Cause: a model = "gpt-5.4" pin. Codex migrates the session to GPT-6 Sol. Recovery: set the model you mean in config.toml, or remove the pin and use the bundled default.

The agent runs commands you expected to approve. Cause: auto mode became the starting mode in v2.1.283, and a classifier now approves actions it judges safe. Since v2.1.285, claude -p and Agent SDK sessions on third-party providers, or with telemetry off, start in auto mode too. Recovery: start sessions with --permission-mode manual, or turn auto mode off organisation-wide with permissions.disableAutoMode: "disable" in managed settings. auto set in project settings is ignored, so a repository cannot switch it on for you.

Claude follows instructions nobody wrote for it. Cause: from v2.1.277 Claude Code loads AGENTS.md when there is no CLAUDE.md, so a file written for Codex now steers Claude too; on 2026-10-12 that holds on both release channels. Recovery: review that AGENTS.md, or add a CLAUDE.md, or switch the behaviour off under Project instructions in /config.

A pinned model is about to be retired. Cause: gpt-5.5 leaves Codex on 2026-10-14, and Claude Sonnet 4.5 retires on 2026-11-30 on Anthropic-operated platforms while Claude Code still names it as the sonnet alias on Bedrock, Google Cloud and Foundry. Recovery: change the pin before the date, name the model you mean in config.toml or your Claude Code settings, and run the repository search from the ten-minute check.

A headless Claude Code run loses an MCP server or a background process. Cause: since v2.1.286 --bare connects only the MCP servers named on the command line, and since v2.1.285 background shell commands stop after their timeout (default 30 minutes, maximum 2 hours). Recovery: name every server the job needs on the command line, and give long-lived commands a longer timeout.

A guide on this site contradicts the tables above. Cause: the guide predates the release. Every page shows its last-updated date; if it is older than the release, trust this page and the per-tool changelog, and report the page so it gets fixed.

Vendor releases in 2026 that changed agent setups

Only changes that alter defaults, remove options or add a model appear here; per-version detail is on the Claude Code changelog, Codex changelog and Cursor changelog pages. Dates marked “secondary” come from press reports rather than the vendor.

October 2026

• Oct 9 · Cursor 3.24. Auto-review stops risky deletes for your approval, admins can restrict Agent to approved remote workspaces with the managed policy AgentRemoteWorkspaceAuthorities, and Claude Opus 5.5 and Sonnet 5.5 run on AWS Bedrock with your own key.
• Oct 9 · Claude Code v2.1.296 (latest). Adds CLAUDE_CODE_WORKFLOW_SUBAGENT_MODEL to run every workflow agent on one model, and prices Sonnet 5.5 cache reads at the lowered rate in /cost, the status line and SDK cost figures.
• Oct 9 · Codex CLI 0.162.1 (stable). Since 0.157.1 no top-level flag or subcommand was added or removed; 0.162.0 (Oct 8) added managed Git worktree tools and /copy.
• Oct 8 · Claude Code v2.1.295 (latest). Command and HTTP hooks gain onFailure, "continue" by default or "block".
• Oct 7 · Anthropic: Claude Haiku 5.5. The haiku alias on the Anthropic API from v2.1.293 (latest channel).
• Oct 7 · Anthropic: monthly API credits for Max and Team plans. They cover the API and the Agent SDK, not interactive Claude Code.
• Oct 7 · Codex CLI 0.161.0. Adds /mcp login <name> and codex exec --cyber-access-program, and restates GPT-6.1 Sol as the default in the bundled and Amazon Bedrock catalogs.
• Oct 6 · Cursor: remote control for local agents. See and reply to the agents on your computer from the Cursor iOS app.
• Oct 6 · Official Terminal-Bench 4.0 leaderboard. Eight rows are added, among them Claude Opus 5.5, Claude Sonnet 5.5 and GPT-6.1 Sol. GPT-6 Astra has had rows since Sep 3, which the 2026-09-26 read of the GitHub submissions folder missed.
• Oct 6 · Anthropic: Claude Mythos 5.1 now requires verification. The Cyber Verification Program grows to three tiers, and Mythos 5.1 is available only to organizations verified through Anthropic’s verification programs, which replaces the invitation-only access in the Sep 1 entry below.
• Oct 5 · Anthropic: the earlier Claude in Slack app is retired in workspaces paired with a Claude organization; Claude Tag replaces it.
• Oct 2 · GitHub Copilot removes four models: Kimi K2.7 Code, Gemini 3.5 Flash, Gemini 3.6 Flash and Claude Opus 4.7. GitHub’s announcement suggests Claude Opus 5.5 for Opus 4.7, not Claude Opus 5 as the cross-tool note in the September section says.
• Oct 2 · Claude Code v2.1.288 (latest). claude project purge becomes claude purge.
• Oct 1 · Claude Code v2.1.287. Adds Claude Mods, plugins that modify deeper behavior, and defaults Opus 4.7 and later and Fable to a 1M window on Bedrock, Google Cloud and Foundry. It is the stable channel on Oct 12.

September 2026

• Sep 30 · Anthropic deprecates Claude Sonnet 4.5. It retires on Nov 30 on Anthropic-operated platforms; Anthropic names claude-sonnet-5-5 as the replacement.
• Sep 29 · OpenAI: GPT-6.1 Sol (DevDay). Becomes Codex’s bundled default in CLI 0.159.1, and GPT-6 Astra moves to second in the picker. Codex CLI 0.159.0 removed the bundled plugin-creator skill.
• Sep 29 · Claude Code v2.1.285. claude -p and Agent SDK sessions on third-party providers, or with telemetry off, start in auto mode; adds the allowedProviders managed setting and claude --desktop.
• Sep 28 · Anthropic: Claude Sonnet 5.5. The sonnet alias on the Anthropic API from Claude Code v2.1.284, where ultracode also becomes its own toggle; Sonnet 5 moves to legacy.
• Sep 28 · Codex CLI 0.158.0. Terminal input approval is on by default, and GPT-5.4 leaves the bundled catalogs.
• Sep 25 · Claude Code v2.1.283. Auto mode becomes the starting permission mode for interactive terminal and VS Code sessions on every plan and provider (latest channel).
• Sep 23 · Codex CLI 0.156.1. GPT-6 Sol and GPT-6 Luna join the model picker, below GPT-6 Astra.
• Sep 22 · Anthropic: Claude Opus 5.5. Becomes Claude Code’s default from v2.1.280; Opus 5 moves to legacy. Anthropic also raised five-hour usage limits on Pro, Max, Team and seat-based Enterprise.
• Sep 21 · SpaceXAI: Grok 4.7. Added to GitHub Copilot the same day; available in Cursor from day one (secondary: MarkTechPost).
• Sep 18 · Claude Code v2.1.277. Reads AGENTS.md when a project has no CLAUDE.md.
• Sep 9 · Codex CLI 0.154.0. Removes codex mcp-server.
• Sep 4 · Codex CLI 0.153.4. GPT-6 Astra becomes the bundled default.
• Sep 2 · Google: Gemini 3.8 Flash (secondary: The Register). In GitHub Copilot from Sep 3.
• Sep 1 · Anthropic: Claude Fable 5.1 and Claude Mythos 5.1. Fable 5.1 replaces Fable 5 in Claude Code from v2.1.257; Mythos 5.1 is invitation-only through Project Glasswing.

August 2026

• Aug 31 · Codex retires GPT-5.4, migrating pinned sessions to GPT-6 Sol.
• Aug 14 · Claude Code auto mode becomes the starting mode on Pro, Max and Team.
• Aug 7 · Codex CLI 0.147.0. Removes codex exec --full-auto; automatic approval review (--approve-for-me) routes approvals through a reviewer model.
• Aug 4 · Claude Code v2.1.222 removes /ultraplan from the command reference; use plan mode (/plan). The same day OWASP published the GenAI LLM Top 10 2026.

July 2026

• Jul 28 · MCP specification 2026-07-28. A stateless core without the initialize handshake; tools adopt it at different speeds (see the cross-tool notes above).
• Jul 24 · Anthropic: Claude Opus 5.
• Jul 9 · GPT-5.6 Sol, Terra and Luna reach GA on Azure (model version 2026-07-09); now labelled “older” in Codex. The Codex desktop app merged into the ChatGPT desktop app as a Codex mode (secondary: Neowin).

June 2026

• Jun 30 · Anthropic: Claude Sonnet 5. Its announced price rise was later cancelled; current prices are on the models hub.
• Jun 9 · Anthropic: Claude Fable 5 and Claude Mythos 5.
• Jun 1 · Claude Code v2.1.160. The dynamic-workflow trigger keyword changes from workflow to ultracode.
• Jun 1 · GitHub Copilot replaces request-based billing with usage-based billing.
• June · Windsurf renamed Devin Desktop (secondary: creeta).

Before June 2026

• May 20 · Cursor 3.5 ships a bundled /loop skill (verified on 2026-08-28).
• May 15 · Roo Code extension shut down, per the Roo-Code README; the repository is archived.

What this guide shipped

Each entry links to the page it describes. Every page also carries its own last-updated date.

Start here, by role Reading paths for developers, tech leads, CTOs and executives.

When a new model ships A 48-hour evaluation playbook before you accept a new default.

The State of Agentic Engineering Sourced evidence on how much code agents write in 2026.

Compound engineering Every's plan, work, review and compound loop, audited.

Vibe coding vs agentic engineering 21 names for working with AI compared: origin, who writes the code, who checks it.

October 2026

• Oct 16 · Code-review guides updated from Anthropic’s 2026-10-15 article. Anthropic’s Claude Code team reported how it layers review on its own repositories (How the Claude Code team does code review; its figures are self-reported). The guides now cover managed Code Review levels, merge gates and costs, a sharper REVIEW.md, answering every finding, review-round caps and blocking security review, with the live Code Review documentation re-checked on 2026-10-16: automated code reviews with Claude Code, security audits with Claude Code, reviewing an agent’s pull request, the review queue, AI code review bots, reading evidence instead of code, governance and autonomy and human in the loop.
• Oct 12 · Docs re-checked against the October releases. Pages that name models and tool versions were re-checked against Claude Code 2.1.296, Codex 0.162.1 and Cursor 3.24, with Claude Sonnet 5.5, Claude Haiku 5.5 and GPT-6.1 Sol covered across the guides and the models hub. The last-updated date on each page tells you whether it has been re-checked.
• Oct 10 · Vibe Coding vs Agentic Engineering: 21 Terms Compared. A free article that defines 21 names for working with AI against vibe coding, each with its origin and a dated, linked quote. The state of agentic engineering now says that Karpathy endorsed the name “agentic engineering” on 4 February 2026 rather than naming it on 30 April 2026, and the A–Z reference gained 16 entries.

September 2026

• Sep 26 · Documentation reorganised by role and lifecycle stage. Role reading paths under Start here for developers, tech leads, CTOs and executives; the method in Method; cross-tool practice in Workflows; models, MCP, skills, plugins and observability under Ecosystem; and new leadership sections, Strategy and business and Engineering organization. Old URLs redirect to their new location.
• Sep 26 · Model facts re-checked. New model pages: new-model evaluation playbook, benchmarks and open-weight models. Pages are re-verified against the installed CLIs as they are rewritten; the last-updated date on each page tells you whether it has been.
• September · The AI-native lifecycle. The lifecycle section maps the artifact chain, plan, design, build, test, deploy, maintain and metrics onto Claude Code, Codex and Cursor.
• September · Human-in-the-loop default changed. Layered agent review plus human gates replaced blanket line-by-line review, which is now reserved for critical and regulated code.
• September · Scorecards rebuilt. The developer scorecard guide and CTO scorecard guide now measure lifecycle maturity.

August 2026

• Aug 28 · The Ladder. Six free hub pages organise the documentation by autonomy level, from assisted coding to running a software factory.
• Aug 28 · The State of Agentic Engineering, August 2026. Third-party evidence, each number with publisher and date, on how much production code agents write, with the counter-evidence sourced the same way.
• Aug 28 · Claude Code routines and Cursor cloud agents and automations. Unattended cloud runs in both tools: triggers, scoping and the untrusted-input rule.
• Aug 24 · Compound engineering. Every’s loop as a cross-tool guide, with an audit of the plugin behind it.
• Aug 20 · The AI Developer Toolkit MCP server. A public, read-only endpoint with search and fetch across every guide on the site.
• Aug 19 · Anti-slop writing skills. unslop and miodkuj, including the deterministic checks that run in CI without an agent.
• Aug 13 · /goal and /loop corrected. Tool support re-verified: Cursor’s /loop skill and Codex goals being on by default had been missed.
• Aug 12 · Software factories and Dr. Skill. When a factory works and when it erodes the codebase; a read-only audit of your skill and MCP loadout.
• Aug 6 · Answer-first rewrite. Every article opens with a paragraph that answers its title, enforced by the build.
• Aug 4 · Pruning context files. An ablation protocol for cutting CLAUDE.md, Cursor rules and AGENTS.md.

July 2026

• Jul 30 · herdr, tmux for agent fleets and agent shells. Running and supervising many agents from one terminal or shell.
• Jul 27 · Claude Opus 5 across the model guides.
• Jul 25 · Mobbin MCP and Impeccable rebuilt for v4.
• Jul 21 · Taste Skill. The anti-slop frontend skill library.
• Jul 11 · Superpowers. A skills plugin that enforces a design-to-PR workflow.
• Jul 3 · Claude Code features you haven’t tried.

June 2026

• Jun 24 · /grill-me and the first cross-tool /goal and /loop guides.
• Jun 10 · Claude Fable 5 covered across the model and routing guides.
• Jun 8 · Official skills from 11 more publishers.
• June · Reducing MCP token cost and skill supply-chain security.

Earlier releases of this guide

• May 2026 · Per-question guides for the developer and CTO scorecards, and bilingual scorecards.
• April 2026 · The four-stage AI design pipeline and the official skills guide.
• February 2026 · A full Codex section alongside Cursor and Claude Code, and Polish translations of every section.
• 17 July 2025 · Version 1.0.0. The first release, covering Cursor and Claude Code.

Where to go next with tool and model changes

• The AI model comparison for coding agents holds the current prices, context windows and defaults this page links to instead of repeating.
• The per-tool changelog pages carry every version: Claude Code, Codex and Cursor.
• The troubleshooting guide covers failures that are not caused by a release.
• The What’s New page shows the newest guides as cards; send feedback when a page lags behind a release.

Frequently Asked Questions

Why does my teammate’s Claude Code session use a different default model?

Claude Code ships on two release channels (latest and stable), and third-party cloud providers set their own model aliases. Compare claude --version and autoUpdatesChannel across machines, and check current channel tags with npm view @anthropic-ai/claude-code dist-tags.

How do I check whether my agent setup is behind the current releases?

Record claude --version and codex --version, run claude doctor and codex doctor, inspect bundled models with codex debug models --bundled, and search the repository for removed flags and legacy model IDs before updating on a branch.

How do I know an agent tool upgrade did not break our workflow?

Pin the CLI versions installed in CI, bump them in a single pull request, and require the test suite, lint, type checks, and a fixed headless smoke task to pass on the new version before merging.