docs: recommend max_completion_tokens

[appleboy]

Summary

Recommend max_completion_tokens: "2000" for new configurations in the English, Traditional Chinese, and Simplified Chinese READMEs. Explain precedence over max_tokens, continued support for legacy services, and automatic conversion for recognized reasoning models when the new input is unset.

Change classification and scope

• Leaf change: documentation only; no runtime changes.
• Goal: document the recommended setting and existing compatibility behavior consistently across all three READMEs.
• Related issues: none supplied; no GitHub or Jira issue linked.

AI authorship

• Tool: OpenAI Codex.
• AI-authored files: README.md, README.zh-TW.md, README.zh-CN.md.
• Human line-by-line reviewed: None — not yet reviewed by a human.

Verification

Prerequisites: Git and a text editor; check out docs/recommend-max-completion-tokens and run commands from the repository root. No services or credentials are required.

1. Inspect the added paragraph immediately after the Inputs table in each README. Expect the same YAML example, precedence rule, legacy-service exception, and conditional reasoning-model conversion in all three languages. Passed: inspected all three additions; the statements and example agree.
2. Compare that guidance with buildChatRequest and reasoningModelPrefixes in main.go. Expect explicit completion tokens to win, recognized reasoning models to convert the legacy field when unset, and other models to retain the legacy field. Passed: inspected the existing implementation and confirmed the documented behavior.
3. Run git diff origin/main...HEAD --check. Expect exit status 0 and no whitespace errors. Passed before commit: git diff --check and git diff --staged --check both exited 0 with no output.

Runtime tests, build, and Go/Docker linters: Not run because only Markdown prose and an example were added; no executable code or build configuration changed. No Markdown-specific check is configured in the repository.

Cleanup: none; inspection does not create data.

Risk and review

• No secrets or security-sensitive interfaces changed.
• Risk: low; inaccurate wording could mislead configuration choices.
• Rollback: revert this documentation commit.
• One reviewer should check the three translations and their agreement with main.go.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.