fix claude 3.7 think enhance prompt problem.

[moqimoqidea]

Context

Fixed the issue that [Prompt Word Enhancement] cladue non-streaming call failed

Implementation

• Due to the current Anthropic's SDK non-streaming Timeout settings, the maximum allowable token for non-streaming calls is 21333.
• If the user configures the Claude model Max Token is greater than this number, the prompt word enhancement and other functions will immediately report an error, the specific error is as follows:

'Error: Streaming is strongly recommended for operations that may take longer than 10 minutes. See https://github.com/anthropics/anthropic-sdk-python#streaming-responses for more details\n    at Anthropic._calculateNonstreamingTimeout (/Users/moqi/Code/Roo-Code/node_modules/@anthropic-ai/sdk/src/core.ts:415:13)\n    at Messages3.create (/Users/moqi/Code/Roo-Code/node_modules/@anthropic-ai/sdk/src/resources/messages/messages.ts:64:46)\n    at AnthropicHandler.completePrompt (/Users/moqi/Code/Roo-Code/src/api/providers/anthropic.ts:224:46)\n    at WecoderHandler.completePrompt (/Users/moqi/Code/Roo-Code/src/api/providers/wecoder.ts:142:18)\n    at singleCompletionHandler (/Users/moqi/Code/Roo-Code/src/utils/single-completion-handler.ts:23:46)\n    at Lh.value (/Users/moqi/Code/Roo-Code/src/core/webview/ClineProvider.ts:1266:38)'

• The typical user optimization prompt is generally no longer than this constant size: ANTHROPIC_DEFAULT_MAX_TOKENS
• In the case of claude non-streaming calls, setting thinking can cause the user to wait for a very long time, In my prompt word enhancement test, using claude 3.7 and keeping think would give the following error:

"TypeError: Cannot read properties of undefined (reading 'find')\n    at AnthropicHandler.completePrompt (/Users/moqi/Code/Roo-Code/src/api/providers/anthropic.ts:236:35)\n    at processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at Lh.value (/Users/moqi/Code/Roo-Code/src/core/webview/ClineProvider.ts:1266:32)"

Screenshots

before:

after:

No errors, successful execution of prompt word enhancements using Claude 3.7 Think

How to Test

Use the Claude 3.7 Think model to perform prompt word enhancements

Get in Touch

moqi

---

Important

Fixes non-streaming call issue for Claude 3.7 in completePrompt() by adjusting token handling and removing thinking parameter.

• Behavior:
  • Fixes non-streaming call failure for Claude 3.7 in completePrompt() in anthropic.ts by setting max_tokens to ANTHROPIC_DEFAULT_MAX_TOKENS and removing thinking parameter.
  • Addresses error when maxTokens exceeds 21333 due to Anthropic's SDK timeout settings.
• Misc:
  • Removes unused maxTokens and thinking from getModel() call in completePrompt().

^{This description was created by for aa70d75. It will automatically update as commits are pushed.}

[changeset-bot]

⚠️ No Changeset found

Latest commit: aa70d75

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

[mrubens]

Thanks!