What does this skill do, and when should you use it?
This skill migrates code, prompts, and API calls from Claude Sonnet 4.0, Sonnet 4.5, or Opus 4.1 to Opus 4.5. It searches the codebase for model strings and API calls, updates platform-specific identifiers, and removes an unsupported 1M-context beta header. By default, it changes model strings only; prompt changes are applied only when the user explicitly requests them or reports a specific issue. Haiku 4.5 models are explicitly out of scope.
Searches the codebase for model strings and API calls; replaces them with Opus 4.5 identifiers for the Anthropic API, AWS Bedrock, Google Vertex AI, and Azure AI Foundry; removes the context-1m-2025-08-07 beta header and adds a note; adds a high effort parameter when requested; summarizes the changes; and tells the user to request further prompt adjustments if Opus 4.5 causes issues.
- A developer maintains an Anthropic API project using Claude Sonnet 4.0 and wants to move to Opus 4.5.
- A team using AWS Bedrock with Sonnet 4.5 or Opus 4.1 needs to update its model identifiers.
- A team using Google Vertex AI or Azure AI Foundry wants to standardize on Opus 4.5.
- A user reports excessive tool triggering, over-engineering, or insufficient code exploration and explicitly requests prompt changes.
- A maintainer needs to find and remove the 1M-context beta header that is unsupported with Opus 4.5.
How do you install this skill?
- Require user confirmation of the change scope and produce a reviewable diff or backup before editing; do not force high effort by default because it may increase cost and latency.
- Add Azure source-model mappings and resolve the conflict between mandatory effort configuration and the reference instruction to apply it only when requested.
- After migration, verify that the deployed API supports the model, beta header, and output_config, then run the project's existing tests.
- The skill does not address Chinese interaction or mainland-China network reachability; availability should be checked separately when relying on Anthropic, AWS, Vertex, or Azure services.
- Local filesystem
The repository README documents Claude Code installation. For example, on macOS/Linux run curl -fsSL https://claude.ai/install.sh | bash; on Windows run irm https://claude.ai/install.ps1 | iex. Then navigate to the project directory and run claude. The source does not document a separate installation command or folder for this skill.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/anthropics/claude-code.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/plugins/claude-opus-4-5-migration/skills/claude-opus-4-5-migration" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Migrate this codebase from Claude Sonnet 4.5 to Opus 4.5, update the model strings, and remove unsupported beta headers.
Run Claude Code in a project containing the relevant code or prompts and use a request such as: “Migrate this codebase from Claude Sonnet 4.5 to Opus 4.5, update the model strings, and remove unsupported beta headers.” To change prompt behavior, describe the specific issue, such as tools being called too frequently. Do not ask it to migrate Haiku models.
What are this skill's strengths and limitations?
- Covers model identifiers for the Anthropic API, AWS Bedrock, Google Vertex AI, and Azure AI Foundry.
- Clearly separates default model-string migration from issue-driven prompt adjustments.
- Handles the beta header documented as unsupported with Opus 4.5.
- Produces a summary of migration changes for review.
- The source does not provide standalone installation instructions for the skill.
- Prompt adjustments depend on the contents of `references/prompt-snippets.md`, which are not included.
- No test suite or migration verification results are provided.
- Haiku 4.5 is unsupported, and the source table does not list source identifiers for Azure AI Foundry.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Opus 4.5 Migration Assistant this page ✓ Anthropic · Official | 40 · Not recommended | ★ 151k | 11d ago | — |
| Cost Mode | 54 · Use with care | ★ 38 | 11d ago | MIT |
| Prompt Master | 53 · Use with care | ★ 14k | 13d ago | MIT |
| Gemini LiveAPI Client Service Skill ✓ Google · Official | 52 · Use with care | ★ 21k | 11d ago | Apache-2.0 |
| 9Router Chat Skill | 47 · Use with care | ★ 31k | 17d ago | MIT |
How did FollowSkills review this skill?
The skill identifies target model strings, platforms, and the Haiku exclusion, but it instructs broad code changes, beta-header removal, and forced high effort without requiring confirmation, previews, backups, or rollback. Cost, data-flow, and external API impact disclosures are incomplete, so points are deducted. No malware, covert exfiltration, or other clear red-line risk is shown.
The workflow is readable and includes Python, TypeScript, and raw API examples. However, Azure source mappings are missing, the workflow's mandatory effort step conflicts with the reference guidance to configure it only when requested, and there is no defined behavior for search scope, unsupported platforms, no matches, or abnormal input. Static review cannot establish execution or key-path coverage, so the score is capped below 10 and reduced further.
The skill states supported source models, target model, migration objects, and the Haiku non-fit range, with several platform variants. It does not define handling for mixed-model repositories, non-Anthropic SDKs, unsupported inputs, or Azure source versions, and provides no Chinese-language or mainland-China environment guidance, so points are deducted.
The name, description, headings, tables, and separate reference files provide a reasonably clear structure, and several behavioral-difference triggers are documented. Installation and dependency notes, compatibility/version range, changelog, maintenance ownership, and update path are absent; license metadata is unknown and the repository license points to commercial terms. Points are deducted accordingly.
For common model-string migrations, the skill gives directly actionable replacements and includes prompt snippets plus a change-summary requirement. However, forced high effort, beta-header handling, and prompt edits require judgment; Azure coverage is incomplete; and no migration verification or testing procedure is provided. Static evidence cannot establish directly usable representative results, so this receives only limited core-task credit.
The files contain concrete model identifiers, beta names, code examples, and conditional instructions, giving some auditability. There are no committed tests, CI coverage, citations, or independent reproduction records for the skill's key claims; compatibility assertions are principally self-authored documentation. The score is therefore low and subject to the static-review cap.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →