Add Claude Opus 4.8 support

[Marve10s]

Summary

• add Claude Opus 4.8 to the Claude provider model list with the documented effort options
• map ultracode to xhigh effort plus the Claude Code ultracode session setting
• update the Claude Agent SDK dependency and Opus alias to use 4.8

Validation

• Not run per request.

---

Note

Medium Risk
Changes the default opus model alias and effort flags passed to Claude Code/SDK across several models; SDK minor-major bump may affect runtime behavior even though scope is provider configuration.

Overview
Adds Claude Opus 4.8 as a built-in model (shown only when Claude Code is v2.1.154+), with shared per-model reasoning effort lists and a new Ultracode option on Opus 4.8.

Ultracode is wired through agent sessions and claude -p text generation as a Claude Code setting (settings.ultracode) while CLI/SDK effort is normalized to xhigh. normalizeClaudeCliEffort now takes the selected model so xhigh stays on Opus 4.8 but maps to max on older models, and max on Sonnet 4.6 maps to high.

The opus slug alias in contracts now resolves to claude-opus-4-8. Provider health checks surface an Opus 4.8 upgrade hint when the CLI is new enough for 4.7 but not 4.8.

Bumps @anthropic-ai/claude-agent-sdk to ^0.3.154 (lockfile picks up platform-specific SDK packages).

^{Reviewed by Cursor Bugbot for commit 625f0c5. Bugbot is set up for automated code reviews on this repo. Configure here.}

Note

Add Claude Opus 4.8 model with ultracode effort mode support

• Adds claude-opus-4-8 as a built-in model with ultracode and ultrathink effort options and a 200k/1m context window selector, gated behind Claude Code version ≥ 2.1.154.
• Extends normalizeClaudeCliEffort to accept a model parameter, with model-specific mappings: ultracode maps to xhigh, xhigh maps to max except for Opus 4.8, and max maps to high for Sonnet 4.6.
• When ultracode effort is selected, both the Claude adapter and text generation layer now set ultracode: true in CLI settings.
• Updates opus alias in model slug aliases to resolve to claude-opus-4-8, and bumps @anthropic-ai/claude-agent-sdk to ^0.3.154.
• Provider status messaging now shows an Opus 4.8-specific upgrade prompt when the installed Claude Code version is below the minimum.

^{Macroscope summarized 625f0c5.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: e2adb4ff-54bf-4033-bffb-e375ecacc625

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a604f0e929

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[macroscopeapp]

Approvability

Verdict: Needs human review

This PR introduces new feature capability (Claude Opus 4.8 model support) with new effort modes and configuration options that affect runtime behavior. Additionally, there is an unresolved review comment about potentially misleading UI options for Sonnet 4.6 effort settings.

^{You can customize Macroscope's approvability policy. Learn more.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 625f0c5. Configure here.}

[cursor]

Sonnet 4.6 "Max" option silently behaves as "High"

Medium Severity

The CLAUDE_EFFORT_OPTIONS.sonnet46 constant adds a new max option that didn't exist before for Sonnet 4.6. However, normalizeClaudeCliEffort maps max to "high" when the model is claude-sonnet-4-6. This means the UI presents "Max" and "High" as two distinct effort choices, but they produce identical behavior. Users selecting "Max" will unknowingly receive "High" effort, which is misleading.

Additional Locations (1)

• apps/server/src/provider/Layers/ClaudeProvider.ts#L299-L302

 

^{Reviewed by Cursor Bugbot for commit 625f0c5. Configure here.}