Skip to content

feat: 支持在 config.json 中配置 ai_max_tokens #116

Description

@xqliu

背景

当前 max_tokens 只能通过环境变量 AI_MAX_TOKENS 配置,默认值为 2000。

用户在使用 Gemini 时遇到问题(#103),返回空响应:

{
  "finish_reason": "length",
  "completion_tokens": 0,
  "prompt_tokens": 11012
}

需要增大 max_tokens,但当前只能通过环境变量设置,不够方便。

当前状态

  • ai_temperature ✅ 已支持在 config.json 配置
  • ai_max_tokens ❌ 只能通过环境变量 AI_MAX_TOKENS 配置

需求

在 config.json 中添加 ai_max_tokens 配置项,与 ai_temperature 保持一致:

{
  "ai_temperature": 0.1,
  "ai_max_tokens": 4000,
  ...
}

实现要点

  1. 在 config.json.example 添加示例
  2. 在配置加载逻辑中读取该字段
  3. 优先级:config.json > 环境变量 > 默认值(2000)
  4. 更新 mcp/client.go 使用配置值

相关 Issue

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions