External/agentic-soc-platform
1
0
Fork 0
mirror of https://github.com/FunnyWolf/agentic-soc-platform.git synced 2026-08-22 13:12:56 +02:00
Code Issues Packages Projects Releases Wiki Activity

delete mem0 and neo4j

Browse Source
This commit is contained in:
rookit
2026-04-01 15:49:29 +08:00
parent a921747211
commit e669cdbfb5
6 changed files with 1 additions and 213 deletions
Download Patch File Download Diff File Expand all files Collapse all files
AGENTS/agent_knowledge.py
-17
View File
@@ -5,10 +5,6 @@ from langchain_core.documents import Document
from Lib.log import logger
from PLUGINS.Embeddings.embeddings_qdrant import embedding_api_singleton_qdrant, SIRP_KNOWLEDGE_COLLECTION
from PLUGINS.Mem0.CONFIG import USE as MEM_ZERO_USE
if MEM_ZERO_USE:
from PLUGINS.Mem0.mem_zero import mem_zero_singleton
class AgentKnowledge(object):
@@ -32,19 +28,6 @@ class AgentKnowledge(object):
if doc.metadata["rerank_score"] >= threshold:
result_all.append(doc.page_content)
if MEM_ZERO_USE:
result = mem_zero_singleton.search_mem(user_id=SIRP_KNOWLEDGE_COLLECTION, query=query, limit=3)
results = result.get("results", [])
relations = result.get("relations", [])
logger.debug(results)
logger.debug(relations)
for one_record in results:
id = one_record.get("id")
rerank_score = one_record.get("rerank_score", 0)
memory = one_record.get("memory", "")
if rerank_score >= threshold:
result_all.append(memory)
results = json.dumps(result_all, ensure_ascii=False)
logger.debug(f"Knowledge search results : {results}")
return results
Docker/Neo4j/docker-compose.yml
-37
View File
@@ -1,37 +0,0 @@
services:
neo4j:
image: docker.1ms.run/neo4j:latest
#image: neo4j:latest
container_name: neo4j-server
restart: always
ports:
# HTTP (浏览器访问 7474)
- "7474:7474"
# Bolt (程序连接 7687)
- "7687:7687"
environment:
# 配置账号密码 (用户名: neo4j, 密码: neo4j-password-for-agentic-soc-platform)
- NEO4J_AUTH=neo4j/neo4j-password-for-agentic-soc-platform
# 绑定地址到 0.0.0.0 以允许外部访问
- NEO4J_server_default__listen__address=0.0.0.0
# 预装核心插件：APOC (实用函数库) 和 GDS (图算法库)
- NEO4J_PLUGINS=["apoc", "graph-data-science"]
# APOC 权限配置：允许访问系统命令及导入导出
- NEO4J_apoc_export_file_enabled=true
- NEO4J_apoc_import_file_enabled=true
- NEO4J_dbms_security_procedures_unrestricted=apoc.*,gds.*
# 内存限制建议 (根据您的机器内存调整)
- NEO4J_server_memory_heap_initial__size=512M
- NEO4J_server_memory_heap_max__size=1G
- NEO4J_server_memory_pagecache_size=512M
volumes:
# 数据持久化
- ./neo4j/data:/data
- ./neo4j/logs:/logs
- ./neo4j/import:/var/lib/neo4j/import
- ./neo4j/plugins:/plugins
Lib/montior.py
+1 -30
View File
@@ -6,25 +6,18 @@ import time
from typing import Callable
from apscheduler.schedulers.background import BackgroundScheduler
from django.contrib.auth.models import User
from Lib.baseplaybook import BasePlaybook
from Lib.moduleengine import ModuleEngine
from Lib.log import logger
from Lib.moduleengine import ModuleEngine
from Lib.playbookloader import PlaybookLoader
from Lib.threadmodulemanager import thread_module_manager
from Lib.xcache import Xcache
from PLUGINS.Embeddings.embeddings_qdrant import embedding_api_singleton_qdrant, SIRP_KNOWLEDGE_COLLECTION
from PLUGINS.Mem0.CONFIG import USE as MEM_ZERO_USE
from PLUGINS.Redis.redis_stream_api import RedisStreamAPI
from PLUGINS.SIRP.sirpapi import Playbook, Knowledge
from PLUGINS.SIRP.sirpmodel import PlaybookJobStatus, KnowledgeAction, PlaybookModel
if MEM_ZERO_USE:
from PLUGINS.Mem0.mem_zero import mem_zero_singleton
ASP_REST_API_TOKEN = "nocoly_token_for_playbook"
class MainMonitor(object):
MainScheduler: BackgroundScheduler
@@ -82,15 +75,6 @@ class MainMonitor(object):
def start(self):
logger.info("Starting background services...")
# add api user
logger.info("Write ASP_TOKEN to cache")
api_usr = User()
api_usr.username = "api_token"
api_usr.is_active = True
Xcache.set_token_user(ASP_REST_API_TOKEN, api_usr, None)
logger.info("Load PlaybookLoader module config")
PlaybookLoader.load_all_playbook_config()
@@ -159,13 +143,6 @@ class MainMonitor(object):
except Exception as E:
logger.exception(E)
try:
if MEM_ZERO_USE:
result = mem_zero_singleton.add_mem(user_id=SIRP_KNOWLEDGE_COLLECTION, run_id=model.rowid, content=payload_content,
metadata={"rowid": model.rowid})
except Exception as E:
logger.exception(E)
model.action = KnowledgeAction.DONE
model.using = True
logger.info(f"Knowledge stored,rowid: {model.rowid}")
@@ -176,12 +153,6 @@ class MainMonitor(object):
except Exception as E:
logger.exception(E)
try:
if MEM_ZERO_USE:
result = mem_zero_singleton.delete_mem(user_id=SIRP_KNOWLEDGE_COLLECTION, run_id=model.rowid)
except Exception as E:
logger.exception(E)
model.action = KnowledgeAction.DONE
model.using = False
logger.info(f"Knowledge removed,rowid: {model.rowid}")
PLUGINS/Mem0/CONFIG.example.py
-3
View File
@@ -1,3 +0,0 @@
# 是否启用记忆插件, mem0会占用较多资源,请根据实际情况选择启用与否
# Whether to enable the memory plugin, Mem0 will consume more resources, please choose to enable or not according to the actual situation
USE = False
PLUGINS/Mem0/__init__.py
View File
PLUGINS/Mem0/mem_zero.py
-126
View File
@@ -1,126 +0,0 @@
import os
from mem0 import Memory
from Lib.configs import BASE_DIR
from Lib.log import logger
from PLUGINS.Embeddings.CONFIG import EMBEDDINGS_SIZE
from PLUGINS.Embeddings.embeddings_qdrant import EmbeddingsAPI
from PLUGINS.LLM.llmapi import LLMAPI
from PLUGINS.Neo4j.CONFIG import NEO4J_URL, NEO4J_PASSWORD, NEO4J_USER
from PLUGINS.Qdrant.qdrant import Qdrant
class MemZero(object):
def __init__(self):
# embeddings
self.embeddings_model = EmbeddingsAPI.get_dense_model()
# llm
llm_api = LLMAPI()
self.llm_model = llm_api.get_model(tag=["fast"])
self.vector_store = Qdrant.get_client()
# your need to download the model from huggingface.co for the first run.
config = {
"reranker": {
"provider": "huggingface",
"config": {
# you need to use Docker/huggingface/download_model.py to download the bge-reranker-v2-m3 model first
"model": os.path.join(BASE_DIR, 'Docker', 'Huggingface', 'bge-reranker-v2-m3'),
"device": "cpu",
"local_files_only": True,
# you can use the online model if you have GPU and internet access
# "model": "BAAI/bge-reranker-v2-m3",
# "device": "cuda",
}
},
"graph_store": {
"provider": "neo4j",
"config": {
"url": NEO4J_URL,
"username": NEO4J_USER,
"password": NEO4J_PASSWORD,
}
},
"vector_store": {
"provider": "qdrant",
"config": {
"collection_name": "knowledge_mem0",
"client": self.vector_store,
"embedding_model_dims": EMBEDDINGS_SIZE,
"on_disk": True,
}
},
"llm": {
"provider": "langchain",
"config": {
"model": self.llm_model,
}
},
"embedder": {
"provider": "langchain",
"config": {
"model": self.embeddings_model,
}
},
}
self.memory = Memory.from_config(config)
logger.info("MemZero initialized successfully.")
def add_mem(self, user_id: str, run_id: str, content: str, metadata: dict):
"""
result = {
"results": vector_store_result,
"relations": graph_result,
}
"""
result = self.memory.add(content, user_id=user_id, run_id=run_id, metadata=metadata)
return result
def search_mem(self, user_id: str, query: str, limit: int = 5, rerank: bool = True):
"""
result = {"results": [{"id": "...", "memory": "...", "score": 0.8, ...}],"relations":[...]}
"""
result = self.memory.siem_search_by_natural_language(
query,
user_id=user_id,
limit=limit,
rerank=rerank,
)
return result
def delete_mem(self, user_id: str, run_id: str):
result = self.memory.delete_all(user_id=user_id, run_id=run_id)
return result
# def demo(self):
# conversation = [
# {"role": "user", "content": "10.198.125.16是安全部门的扫描器,可能会产生NDR告警,直接忽略"},
# {"role": "user", "content": "test@gmail.com是钓鱼模拟邮箱,如果用户上报的钓鱼邮件是这个邮箱,直接降低等级"},
# ]
#
# result_add = self.memory.add(conversation, user_id="demo-user")
# print(result_add)
# print(time.time())
# results = self.memory.search(
# "test@gmail.com需要安全部门封禁吗?",
# user_id="demo-user",
# limit=3,
# rerank=True,
# )
# for hit in results["results"]:
# print(hit)
# print(time.time())
#
# def delete(self):
# result_delete = self.memory.delete_all(user_id="demo-user")
# print(result_delete)
mem_zero_singleton = MemZero()