克隆插桩库、智能体开发框架和智能体，这三类项目的代码库到本地进行调研。

1. 是否进行了插桩？
   1. 是的话，如何插桩的？
   2. 采集哪些信息？
2. 是否对大模型消息进行了采集
   1. 是的话，如何采集的？
   2. 将此消息写入到什么类型的遥测里了？logs还是traces？具体是哪个字段？

Make a research and write the report into markdown file. Question 2 and 3 is for agent frameworks and agent proejcts. Question 1 is for OpenTelemetry's projects:
1. What is the latest progress of agent observability according to OTel's repo(community, semantic-conventions and opentelemetry-specification)? 
   3.1 Does OTel suggest any instrumentation library?
   3.2 What information OTel suggest to collect?
   3.3 Does OTel suggest collecot the LLM messages?
   3.4 Which field should be store the LLM messages?
   3.5 Any thing with observability 2.0?
2. If this project has been instrumented (not for instrumentation library)
   2.1 If so, how instrumentation implemented. Using what library?
   2.2 Collect what information?
   2.3 Store them into what kind of telemetry? traces, metris, logs, wide event(observability 2.0) or their own data structure.
   2.4 Does instrumentation follow the best practices or standard designed by OpenTelemetry
3. If this project collect the LLM messages
   3.1 If so, how LLM messages been collected?
   3.2 Which kind of telemetry store the LLM messages? traces, metris, logs, wide event(observability 2.0) or their own data structure.
   3.3 Which field the LLM message has been write down?
   3.4 If this instrumentation follow the best practices or standard designed by OpenTelemetry