[ENHANCEMENT] Added customizable Anthropic options. May have fixed some hidden performance issues (uncertain).

[ygzzfyh123]

Problem (one or two sentences)

1. The Anthropic protocol does not allow customization of the model context window and price.

2. Sometimes the plugin may crash; see " [BUG] Chat history disappeared AND Plugin service crashes frequently #715 "[BUG] Chat history disappeared AND Plugin service crashes frequently #715 issues

Context (who is affected and when)

1. When setting up the model

Desired behavior (conceptual, not technical)

This is an issue used to submit pull requests.

Constraints / preferences (optional)

No response

Request checklist

• I've searched existing Issues and Discussions for duplicates
• This describes a specific problem with clear context and impact

Zoo Code Task Links (optional)

No response

Acceptance criteria (optional)

No response

Proposed approach (optional)

No response

Trade-offs / risks (optional)

No response

[edelauna]

Could you provide more details as to why this is required?

[ygzzfyh123]

Could you provide more details as to why this is required?

Reasons for creating this PR:

The main reason for opening this PR is to address compatibility issues caused by certain third-party API vendors and to provide a more flexible configuration option. The specific reasons are as follows:

Protocol Compatibility and Errors: Some third-party API providers have poor support for Claude and protocol conversion. Using OpenAI-compatible protocols with them results in errors indicating that tools cannot be called.

Context Length Limitations: Although switching to the Anthropic protocol resolves the tool-calling errors, the default Anthropic protocol does not support custom context lengths.

Non-Destructive Solution: To resolve this dilemma, this PR adds a new option in a non-destructive manner. It does not directly modify the original Anthropic implementation, ensuring no negative impact while perfectly adapting to the limitations of third-party APIs.