Feat: Add Anthropic CUA adaptive thinking

[chromiebot]

why

what changed

test plan

---

Summary by cubic

Add adaptive thinking for Anthropic Claude 4.6 models in the CUA client. Uses thinkingEffort with output_config.effort, sets temperature=1, and keeps thinkingBudget for older models.

• New Features

  • Detects 4.6 models (claude-opus-4-6, claude-sonnet-4-6) and uses thinking: { type: "adaptive" } with output_config.effort.
  • Defaults to "medium" effort when not set and forces temperature: 1 for adaptive thinking.
  • Adds ThinkingEffort and thinkingEffort in ClientOptions ("low" | "medium" | "high" | "max"); falls back to thinking: { type: "enabled", budget_tokens } with thinkingBudget on older models.
  • Uses computer_20251124 for 4.6 and claude-opus-4-5-20251101; tests cover effort levels, defaults and temperature, provider-prefixed names, and legacy budget behavior.

• Migration

  • For 4.6 models, set thinkingEffort; thinkingBudget is deprecated. Adaptive thinking sets temperature: 1 automatically.
  • Keep using thinkingBudget on older Claude models. No changes needed if you don’t use thinking.

^{Written for commit 0ea0332. Summary will update on new commits. Review in cubic}

[changeset-bot]

⚠️ No Changeset found

Latest commit: 0ea0332

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[github-actions]

This PR is from an external contributor and must be approved by a stagehand team member with write access before CI can run.
Approving the latest commit mirrors it into an internal PR owned by the approver.
If new commits are pushed later, the internal PR stays open but is marked stale until someone approves the latest external commit and refreshes it.

[cubic-dev-ai]

No issues found across 3 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

Architecture diagram

sequenceDiagram
    participant App as Application/User
    participant Client as AnthropicCUAClient
    participant SDK as Anthropic SDK (Beta)
    participant API as Anthropic API

    Note over App,API: Request Flow for Claude 4.6+ (Adaptive Thinking)

    App->>Client: getAction(messages, options)
    Note right of Client: Includes NEW: thinkingEffort (low/medium/high/max)

    Client->>Client: Identify Model Version
    
    alt Model is Claude 4.6+ (Opus/Sonnet)
        Client->>Client: NEW: Set thinking.type = "adaptive"
        Client->>Client: NEW: Set output_config.effort = thinkingEffort
        Client->>Client: NEW: Select computer_20251124 tool
    else Model is Legacy (Claude 4.5 or older)
        Client->>Client: CHANGED: Set thinking.type = "enabled"
        Client->>Client: CHANGED: Set thinking.budget_tokens = thinkingBudget
        Client->>Client: Select legacy computer tool version
    end

    Client->>SDK: beta.messages.create(payload)
    Note right of SDK: Payload contains specific<br/>thinking config & tool versions

    SDK->>API: POST /v1/messages (Beta Headers)
    API-->>SDK: Response with thinking block + tool use
    SDK-->>Client: Message object
    Client-->>App: Action/Response

Loading

[cubic-dev-ai]

1 issue found across 2 files (changes from recent commits).

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="packages/core/tests/unit/anthropic-cua-adaptive-thinking.test.ts">

<violation number="1" location="packages/core/tests/unit/anthropic-cua-adaptive-thinking.test.ts:179">
P3: Duplicate coverage: the new claude-sonnet-4-6 temperature test repeats the existing low-effort adaptive-thinking test with identical setup and assertions, adding no new behavior coverage.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[cubic-dev-ai]

P3: Duplicate coverage: the new claude-sonnet-4-6 temperature test repeats the existing low-effort adaptive-thinking test with identical setup and assertions, adding no new behavior coverage.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At packages/core/tests/unit/anthropic-cua-adaptive-thinking.test.ts, line 179:

<comment>Duplicate coverage: the new claude-sonnet-4-6 temperature test repeats the existing low-effort adaptive-thinking test with identical setup and assertions, adding no new behavior coverage.</comment>

<file context>
@@ -142,9 +146,57 @@ describe("AnthropicCUAClient adaptive thinking", () => {
+      );
+    });
+
+    it("should set temperature to 1 for claude-sonnet-4-6 with adaptive thinking", async () => {
+      const client = new AnthropicCUAClient(
+        "anthropic",
</file context>

[github-actions]

This PR was approved by @miguelg719 and mirrored to #1954. All further discussion should happen on that PR.