Update Models and Agent CLIs

Refresh the default model selections in this dotfiles repo using the latest official release information from Anthropic, OpenAI, and Google. Update the Claude Code, Codex, and Gemini CLIs through their configured managers, then sweep the invoking repository for any hardcoded model IDs that should be bumped too.

Targets

Invocation: $my-update-models [$provider].

Treat an omitted $provider as all. Every target always includes its model settings, CLI, and the invoking repository's model-ID scan.

• claude — Anthropic model settings and native Claude Code
• codex — OpenAI model settings and mise-managed Codex CLI
• gemini — Google model settings and mise-managed Gemini CLI
• all — all three model settings and CLIs

Reject unknown providers before making network calls.

Examples:

$my-update-models          # all model settings, CLIs, and the repo scan
$my-update-models codex    # Codex model settings, CLI, and the repo scan
$my-update-models gemini   # Gemini model settings, CLI, and the repo scan

Dotfiles Config Files

All paths are relative to /home/yuki/ghq/github.com/Ynakatsuka/dotfiles/. Always edit in the ghq repo, never in ~/.

File	Field	Example value	Notes
home/.chezmoitemplates/claude-settings.json	top-level model	"opus"	Default Claude Code model alias. Effort is managed separately via the top-level effortLevel.
home/.chezmoitemplates/claude-settings.json	env.CLAUDE_CODE_SUBAGENT_MODEL	"opus"	Subagent model alias. Only present on some setups — skip if the key is absent.
home/.chezmoitemplates/claude-settings.json	autoUpdatesChannel	"latest"	Native Claude Code update channel. Read it when checking or updating the CLI; do not change it unless requested.
home/dot_codex/private_config.toml.tmpl	model	"gpt-6-sol"	Codex CLI default model (full ID, not an alias).
home/.chezmoitemplates/gemini-settings.json	model.name	"pro"	Accepts aliases (auto, pro, flash, flash-lite) or full IDs (e.g. gemini-2.5-pro). Aliases auto-track the CLI default across releases — keep the alias unless the user wants a pinned version.

Do NOT modify model IDs that appear inside skill examples (e.g., home/dot_claude/skills/my-agent/SKILL.md). Those are illustrative only.

CLI Management

Keep the configured split ownership. Do not migrate a CLI to another manager as part of this skill.

CLI	Expected manager	Current version	Latest/install/update command
Codex	mise entry npm:@openai/codex = "latest"	codex --version	Latest: mise latest npm:@openai/codex; install: mise install npm:@openai/codex@latest; update: mise upgrade npm:@openai/codex --yes
Gemini	mise entry npm:@google/gemini-cli = "latest"	gemini --version	Latest: mise latest npm:@google/gemini-cli; install: mise install npm:@google/gemini-cli@latest; update: mise upgrade npm:@google/gemini-cli --yes
Claude Code	Anthropic native installer under ~/.local/share/claude/versions/	claude --version	Check the configured channel and official releases; update with claude update

For Codex and Gemini, first verify the expected mise entry and inspect the installed state:

mise ls --json npm:@openai/codex
mise ls --json npm:@google/gemini-cli

If the selected package is configured but not installed, report its installed version as not installed and propose the table's mise install command. This is not an ownership mismatch. For an installed mise package, resolve the executable with command -v, inspect the symlink with realpath, and run its version command. Resolve Claude Code the same way before proposing an update.

Stop and report an ownership mismatch if the configured manager or resolved executable differs from the table. Do not fall back to npm install -g, curl | sh, or another installer. For Claude Code, also stop if DISABLE_UPDATES=1 prevents manual updates.

Primary Sources

Use primary sources only. If a required primary source is unavailable, report the failure and stop before proposing or applying an update.

Anthropic (Claude / Claude Code)

• Models overview: https://docs.anthropic.com/en/docs/about-claude/models/overview
• API pricing: https://platform.claude.com/docs/en/about-claude/pricing
• News: https://www.anthropic.com/news
• Claude Code setup and updates: https://docs.anthropic.com/en/docs/claude-code/setup
• Claude Code release notes: https://docs.claude.com/en/release-notes/claude-code
• Claude Code releases: https://github.com/anthropics/claude-code/releases

OpenAI (Codex)

• Models reference: https://platform.openai.com/docs/models
• API pricing: https://developers.openai.com/api/docs/pricing
• Prompt caching: https://developers.openai.com/api/docs/guides/prompt-caching
• News: https://openai.com/news
• Codex CLI repo (releases / changelog): https://github.com/openai/codex

Google (Gemini)

• Models reference: https://ai.google.dev/gemini-api/docs/models
• API pricing: https://ai.google.dev/gemini-api/docs/pricing
• Context caching: https://ai.google.dev/gemini-api/docs/caching
• Gemini CLI installation and release channels: https://github.com/google-gemini/gemini-cli/blob/main/docs/get-started/installation.mdx
• Gemini CLI releases: https://github.com/google-gemini/gemini-cli/releases
• Gemini CLI default + alias table: https://github.com/google-gemini/gemini-cli/blob/main/packages/core/src/config/models.ts

Workflow

1. Parse scope. Resolve $provider using the default and validation rules above. Report the selected provider(s), model settings, CLIs, and invoking repository before continuing.
2. Read current state. Open the managed model settings for the selected provider(s), verify each selected CLI's ownership, and capture the current model values and installed CLI versions. Report the current state up front.
3. Fetch latest info. Use WebFetch on the primary sources. If a required source cannot be read, report the failure and stop before proposing an update. Fetch only information required by the selected provider(s):
   • Newest available model IDs (new Claude family generation, new gpt-*-codex release, new gemini-* generation)
   • Release date and any deprecation notice on the currently configured model
   • Whether a new alias has been introduced (e.g., a new Claude family, or a new Gemini alias)
   • When a model update is available, the current and proposed models' API prices, including every published cache write, cache read/hit, and cache storage rate
   • Latest selected CLI versions. Use mise latest npm:@openai/codex for Codex and mise latest npm:@google/gemini-cli for Gemini. For Claude Code on the latest channel, use the official GitHub release and cross-check it against the changelog. For another channel, report the channel explicitly and use only a target documented for it. The mise-managed Gemini entry tracks npm's latest dist-tag, so compare it with the newest stable, non-prerelease GitHub release and report any mismatch.
4. Compare models. Build a short table: file | field | current | proposed | reason. Note trade-offs (capability, latency, cost) when relevant.
5. Compare prices when an update is available. Before asking for approval, show the pricing table defined below. Include adjacent current and proposed rows for every model change. Omit this table only when no model update is available.
6. Compare CLIs. Build a short table: CLI | manager | installed | latest | action.
7. Confirm before mutating anything. Let the user independently approve model config edits, each CLI update, and repo-scan edits. If the user picks a different choice, follow it.
8. Update selected CLIs. Run only the approved install or update command from the CLI Management table. For a mise-managed CLI, use mise install when it is configured but absent and mise upgrade when it is installed. Run mise reshim --force once after any successful mise install or upgrade. If a command fails, surface the error and stop that update; do not switch installers or claim partial success as complete.
9. Verify CLI updates. Re-resolve each updated executable, run its --version command, and compare the result with the proposed version. Report an error if the executable moved to an unexpected manager or the installed version did not change as expected.
10. Apply approved model edits. Edit the ghq repo with the Edit tool. Change one field per edit.
11. Update and run the tests for edited config. Tests in scripts/ assert managed defaults, so a model edit is incomplete until they match. For each edited file and old value, list the tests that read the file or pin the value, then update only assertions of the managed default in the same change:

rg -l -F -e '<edited file path>' -e '<old value>' scripts/

Leave fixtures that deliberately use an older ID. Run each affected test by name with the repository test runner, (cd scripts && bash test.sh <test-name>); if one fails, report its output and stop before deploying. 12. Scan the invoking repository. Scan for hardcoded model IDs as described below. 13. Deploy model config changes. Tell the user the deploy steps; only run them if asked. The chezmoi source dir is the ghq clone, so the canonical sequence is:

chezmoi git pull -- --ff-only
chezmoi diff
chezmoi apply -v

Per repo policy, do not commit automatically — wait for explicit approval. 14. Verify model config changes. After chezmoi apply, read ~/.codex/config.toml, ~/.claude/settings.json, and ~/.gemini/settings.json to confirm the change landed.

Model Price Comparison

When at least one configured model has a newer proposed model, ALWAYS show both the current and proposed models in this exact table shape:

Provider	Role	Model	Input	Cache write	Cache read/hit	Cache storage	Output	Pricing basis
...	current	...	...	...	...	...	...	...
...	proposed	...	...	...	...	...	...	...

Use the public paid, standard, on-demand API tier as the common comparison basis unless the configured model is available only through another documented tier. State the currency and units, normally USD per 1M tokens; include per hour for cache storage. Preserve context-length bands and other pricing thresholds instead of choosing the cheapest row. If the CLI or subscription plan bills differently from the API, say that the API prices are comparison figures and do not represent the subscription charge.

Never leave a cache column blank and never use an unexplained dash. Copy every applicable cache rate from the provider's official pricing page:

• Include all cache-write durations when they have different rates, such as 5-minute and 1-hour writes.
• Include both the cached-input/read rate and token-hour storage rate when the provider prices them separately.
• If the official pricing scheme has no separate charge for a cache operation, write not separately charged and cite the official source.
• If the provider does not publish any input, cache, storage, or output rate, write not published (checked YYYY-MM-DD) in that cell. Do not substitute a sibling model's price or infer a value.

Put the official pricing link in Pricing basis for each row. Use the same pricing basis for the current and proposed rows so the comparison is valid.

Repo Scan (hardcoded model IDs)

After updating dotfiles configs, sweep the current working directory's repository (the one the user invoked the skill from) for hardcoded model IDs that also need bumping.

1. Determine the repo root.

   git rev-parse --show-toplevel
   

   If the command fails (not a git repo), skip the scan and tell the user.

2. Grep for model-ID patterns. Run the Grep tool at the repo root with this regex (ripgrep syntax, respects .gitignore by default):

   claude-(fable|opus|sonnet|haiku)-[0-9]|gpt-[0-9]|gemini-[0-9]|\bo[134](-mini|-preview)?\b
   

   Prioritize these globs on the first pass:

   • .github/workflows/**/*.{yml,yaml} — GitHub Actions
   • **/*.py — Python code
   • **/*.{sh,bash,zsh} — shell scripts
   • **/*.{ts,tsx,js,jsx,mjs} — Node / TypeScript clients
   • **/*.{toml,json,yaml,yml} — app configs (skip lockfiles)

3. Classify each hit before proposing edits:

   • Update candidate: production code, CI workflow, deployment config, CLI wrapper, Dockerfile.
   • Leave alone:
     • documentation examples, changelog entries, migration notes
     • test fixtures, recorded cassettes, VCR tapes, golden snapshots, except assertions of a default that this run changed (handle those as update candidates)
     • lockfiles, vendor dirs (node_modules/, .venv/, dist/, build/)
     • files under .claude/skills/ or home/dot_claude/skills/ that only illustrate a model name
     • comments that intentionally name a legacy model for historical reasons
     • the dotfiles config files already handled in the previous section

4. Present candidates. Show the user a table: path:line | current | proposed | category. List skipped hits on a separate short line (one reason each).

5. Apply after confirmation. Edit one occurrence at a time. If the same model ID appears many times in a single file with identical context, use replace_all; otherwise edit per-site to preserve surrounding context.

6. Re-run the grep after edits to confirm no stale IDs remain in the candidates you chose to update.

Notes

• Claude Code uses friendly aliases (opus, sonnet, haiku). Anthropic rotates the underlying model behind each alias on release, so the alias itself usually does not need to change. Switch the alias only when picking a different family (e.g., dropping from opus to sonnet for cost) or when a new family ships.
• Codex uses full model IDs. Always update to a specific ID announced on the OpenAI models page or the Codex CLI release notes.
• Gemini CLI accepts both aliases (auto, pro, flash, flash-lite) and full IDs (gemini-2.5-pro, gemini-3-pro-preview). Aliases track the CLI default across releases — keep the alias unless the user wants pinning.
• If the new model has different reasoning controls, also re-evaluate model_reasoning_effort in home/dot_codex/private_config.toml.tmpl and the [1m] effort suffix on the Claude Code model field.