无法加入记忆

[Jacky-Yang123]

您好，我多次尝试类似示例代码无法加入记忆，每次都会自己结束，Managers0 Triggered，偶尔有几次Managers2 Triggered，但是最后也是No updates recorded：

result = client.add(
user_id=user_id,
messages=[
{
"role": "user",
"content": [{
"type": "text",
"text": "MirixDB is our company's new internal database system that combines PostgreSQL with Redis caching. It's specifically designed for handling multi-agent memory operations with automatic cache invalidation and vector search capabilities. We're using it in our Q4 AI assistant rollout."
}]
},
{
"role": "assistant",
"content": [{
"type": "text",
"text": "I've saved this information about MirixDB to my semantic memory."
}]
}
],
chaining=False,
filter_tags=filter_tags,
occurred_at="2025-11-16T10:30:00"
)

但是总是无法更新记忆到Memory Components的这几个分类记忆里
Refresh
Episodic
Semantic
Procedural
Resource
Knowledge
Core

我查看了Memory Traces，发现每次都会自己结束，Managers0 Triggered，偶尔有几次Managers2 Triggered，但是最后也是No updates recorded.

这是Managers0 Triggered部分回复：
2025/12/30 17:33:32
completed
N/A
1 Message(s)
mqt_OUHJ

Execution Flow
meta_memory_agent
completed
Type
meta_memory_agent
Status
completed
Duration
37.8s
Managers
0 Triggered
Assistant Output
"I'll analyze this message about MirixDB and determine which memories need to be updated. Let me think through this step by step: Analysis of the Message: - MirixDB is described as a new internal database system - It combines PostgreSQL with Redis caching - Designed for handling multi-agent memory operations - Has automatic cache invalidation and vector search capabilities - Being used in Q4 AI assistant rollout Memory Classification: 1. Semantic Memory: This contains general knowledge about concepts, systems, and technologies. MirixDB is a NEW concept/system that I didn't previously know about - it's a database technology with specific capabilities. 2. Knowledge Memory: This contains static reference data like technical specifications. The description includes specific technical details (PostgreSQL + Redis, multi-agent memory operations, cache invalidation, vector search) that are factual reference data. 3. Resource Memory: This could potentially include references to documents/files if there are any mentioned, but in this case there are no specific files or documents referenced. The message describes a technical system with specific capabilities, which fits best under semantic memory (understanding what MirixDB is) and knowledge memory (the technical specifications as reference data). Let me trigger the memory updates:<｜tool▁calls▁begin｜><｜tool▁call▁begin｜>trigger_memory_update<｜tool▁sep｜>{"memory_types": ["semantic", "knowledge"]}<｜tool▁call▁end｜><｜tool▁calls▁end｜>"

}
Response
I'll analyze this message about MirixDB and determine which memories need to be updated. Let me think through this step by step:

Analysis of the Message:

• MirixDB is described as a new internal database system
• It combines PostgreSQL with Redis caching
• Designed for handling multi-agent memory operations
• Has automatic cache invalidation and vector search capabilities
• Being used in Q4 AI assistant rollout

Memory Classification:

1. Semantic Memory: This contains general knowledge about concepts, systems, and technologies. MirixDB is a NEW concept/system that I didn't previously know about - it's a database technology with specific capabilities.

2. Knowledge Memory: This contains static reference data like technical specifications. The description includes specific technical details (PostgreSQL + Redis, multi-agent memory operations, cache invalidation, vector search) that are factual reference data.

3. Resource Memory: This could potentially include references to documents/files if there are any mentioned, but in this case there are no specific files or documents referenced.

The message describes a technical system with specific capabilities, which fits best under semantic memory (understanding what MirixDB is) and knowledge memory (the technical specifications as reference data).

Let me trigger the memory updates:<｜tool▁calls▁begin｜><｜tool▁call▁begin｜>trigger_memory_update<｜tool▁sep｜>{"memory_types": ["semantic", "knowledge"]}<｜tool▁call▁end｜><｜tool▁calls▁end｜>

这是triggered2 最后Parallel Memory Updates部分回复

semantic_memory_agent
Type
semantic_memory_agent
Status
completed
Duration
12.2s
Managers
N/A
Updates Summary
No updates recorded.
Assistant Output
2025/12/30 17:24:17
I'll analyze the user's message about MirixDB and add this new concept to semantic memory since it appears to be a new company-specific database system that I'm not already familiar with.<｜tool▁calls▁begin｜><｜tool▁call▁begin｜>semantic_memory_insert<｜tool▁sep｜>{"items": [{"name": "MirixDB", "summary": "MirixDB is our company's new internal database system that combines PostgreSQL with Redis caching, specifically designed for handling multi-agent memory operations with automatic cache invalidation and vector search capabilities.", "details": "MirixDB is a proprietary database system developed internally at our company. It integrates PostgreSQL as the primary database with Redis caching layer to optimize performance for multi-agent memory operations. Key features include automatic cache invalidation mechanisms and built-in vector search capabilities. The system is being deployed as part of the Q4 AI assistant rollout project.", "source": "User message describing company's internal database system", "tree_path": ["work", "projects", "ai-assistant-rollout", "database-systems"]}]}<｜tool▁call▁end｜><｜tool▁calls▁end｜>
Tool Calls (2)
retrieve_memories
26ms
completed
Arguments
{
"topics": null
}
Response
{
"key_words": "",
"core": "",
"knowledge": {
"total_number_of_items": 0,
"current_count": 0,
"text": ""
},
"episodic": {
"total_number_of_items": 0,
"recent_count": 0,
"relevant_count": 0,
"recent_episodic_memory": "",
"relevant_episodic_memory": ""
},
"resource": {
"total_number_of_items": 0,
"current_count": 0,
"text": ""
},
"procedural": {
"total_number_of_items": 0,
"current_count": 0,
"text": ""
},
"semantic": {
"total_number_of_items": 0,
"current_count": 0,
"text": ""
}
}
LLM REQUEST
Response
I'll analyze the user's message about MirixDB and add this new concept to semantic memory since it appears to be a new company-specific database system that I'm not already familiar with.<｜tool▁calls▁begin｜><｜tool▁call▁begin｜>semantic_memory_insert<｜tool▁sep｜>{"items": [{"name": "MirixDB", "summary": "MirixDB is our company's new internal database system that combines PostgreSQL with Redis caching, specifically designed for handling multi-agent memory operations with automatic cache invalidation and vector search capabilities.", "details": "MirixDB is a proprietary database system developed internally at our company. It integrates PostgreSQL as the primary database with Redis caching layer to optimize performance for multi-agent memory operations. Key features include automatic cache invalidation mechanisms and built-in vector search capabilities. The system is being deployed as part of the Q4 AI assistant rollout project.", "source": "User message describing company's internal database system", "tree_path": ["work", "projects", "ai-assistant-rollout", "database-systems"]}]}<｜tool▁call▁end｜><｜tool▁calls▁end｜>

[Jacky-Yang123]

我试着改了下模型和提示词，好像有所改善。我再持续测试下。

[Mirix-AI]

你用的是哪一个模型呀，这个看起来像是meta memory manager想trigger但是sub agent 拒绝更新，有点奇怪

[Jacky-Yang123]

lm studio的qwen3-30b-a3b。今天重新拉了下项目又报requests.exceptions.HTTPError: API request failed: Invalid LLM model: Model 'qwen3-30b-a3b' is not supported. Available models: gemini-2.0-flash, gemini-2.0-flash-lite, gemini-3-flash-preview, gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, gpt-4o, gpt-4o-mini, gpt-5, gpt-5-mini... and 18 more，在哪能去除这个限制吗谢谢。

[Jacky-Yang123]

刚尝试只把模型名随便换成一个支持的如gpt-5又可以正常运行，但是lmstudio后台显示还是调用的qwen3-30b-a3b。虽然过程不对但是能跑了😄记忆目前试了一次也可以正常加入。
此外还有个问题，看lmstudio后台显示会这样重复调用，不过过一会儿就会跳出来正常运行：
2025-12-31 11:26:44 [INFO]
[qwen3-30b-a3b] Tool name generated: update_topic
2025-12-31 11:26:45 [INFO]
[qwen3-30b-a3b] Model generated a tool call.
2025-12-31 11:26:46 [INFO]
[qwen3-30b-a3b] Start to generate a tool call...
2025-12-31 11:26:46 [INFO]
[qwen3-30b-a3b] Tool name generated: update_topic
2025-12-31 11:26:48 [INFO]
[qwen3-30b-a3b] Model generated a tool call.
。。。

[wangyu-ustc]

诶这个不知道是不是lmstudio显示的问题，你可以在mirix dashboard里面看一下嘛；
然后上面这个限制我已经在最新的commit里面去除了