docs: update roadmap

This commit is contained in:
Simon
2025-10-11 17:27:38 +08:00
parent 4585a66c52
commit 23a69e34b2
2 changed files with 2 additions and 38 deletions

View File

@@ -20,6 +20,8 @@ The development progress and future plans for PageAgent.
- [ ] **Working homepage with live LLM API**
- [ ] **free CDN**
- [ ] **Testing suits**
- [ ] **Remove ai-sdk** - Only one function is being used
- [ ] **Support custom llm fetch**
♻️ Following browser-use's update and contribute back
@@ -32,6 +34,4 @@ The development progress and future plans for PageAgent.
## 🤔 To Be Decided
- [ ] **Remove ai-sdk?** - Only one function is being used
- [ ] **Cross-origin multi-page?** - Tricky
- [ ] **Support custom llm fetch?**

View File

@@ -218,42 +218,6 @@ export class PageAgent extends EventTarget {
* - next_goal: string
* - action: { toolName: toolInput }
* where action must be selected from tools defined in this.tools
*
* @topic 要不要合并成一个 tool
* @facts
* - 我们需要模型每步返回 evaluation/memory/goal 等思考过程
* - browser use 合并成一个巨大的 tool
* ```json
* {
* "memory": "...",
* "goal": "...",
* "actions": [
* {
* "name": "...",
* "args": "..."
* }
* // ...
* ]
* }
* ```
* - qwen 目前必须指定 function name 来确保 tool call
* @reasoning
* - 不能为了 qwen 的缺陷而设计系统
* - 更复杂的 tool 更容易出错
* - 分散的 tool 更容易利用 ai-sdk 的重试机制,也更容易处理错误
* - 不能用额外的步骤生成这些数据,不仅性能过差,而且 goal 之类的必须和 call 一起生成
* @options
* - Plan @A
* - 和 browser use 使用完全一致的做法,合并成一个大 tool要求每次调用
* - 会把 tool 定义变得非常复杂,增加出错率
* - Plan @B
* - 每次调用两个 tool其中一个用来输出思考
* - 很难用提示词 enforce 这么复杂的规则
* - Plan @C
* - 自动为每个 tool 增加固定的 reasoning/memory/goal 等输入,并自动拦截提取这些数据
* - 会让 tool 定义变得很长
* @conclusion
* - 使用 @A
*/
#packMacroTool(): ToolSet {
const tools = this.tools