DB-GPT V0.7.0版本更新：支持MCP协议、集成DeepSeek R1模型、GraphRAG检索链路增强、架构全面升级等

DB-GPT
是一款开源的AI原生数据应用开发框架(AI Native Data App Development framework with AWEL and Agents)，在V0.7.0
版本中，我们对DB-GPT
模块包进行架构治理，将原有模块包进行分拆，重构了整个框架配置体系，提供了更加清晰，更加灵活，更加可扩展的围绕大模型构建AI原生数据应用管理与开发能力。

 新特性

支持
MCP(Model Context Protocol)
协议

npx -y supergateway --stdio "uvx mcp-server-fetch"

这里我们运行网页抓取的 mcp-server-fetch

b. 创建  Multi-agent
+ Auto-Planning
+ MCP
 的网页抓取和总结 APP。 

c. 配置 APP，选择 ToolExpert
和 Summarizer
这两个智能体，并给  ToolExpert
 添加资源类型是 tool(mcp(sse))
的资源，其中 mcp_servers
填步骤 a 中启动的服务地址（默认是： http://127.0.0.1:8000/sse
)，然后保存应用。

d. 选择刚才新建的 MCP Web Fetch
APP 对话，给出网页让 APP 总结即可：

例子中输入的问题是：这个网页讲了什么 https://www.cnblogs.com/fnng/p/18744210

[rag.storage.graph]
type = "TuGraph"
host="127.0.0.1"
port=7687
username="admin"
password="73@TuGraph"


enable_summary="True"
triplet_graph_enabled="True"
document_graph_enabled="True"


# 向量图检索配置项
enable_similarity_search="True"
knowledge_graph_embedding_batch_size=20
similarity_search_topk=5
extract_score_threshold=0.7

[rag.storage.graph]
type = "TuGraph"
host="127.0.0.1"
port=7687
username="admin"
password="73@TuGraph"


enable_summary="True"
triplet_graph_enabled="True"
document_graph_enabled="True"


# 意图识别+Text2GQL图检索配置项
enable_text_search="True"


# 使用Ollama部署独立text2gql模型，开启以下配置项
# text2gql_model_enabled="True"
# text2gql_model_name="tugraph/CodeLlama-7b-Cypher-hf:latest"

[system]
# Load language from environment variable(It is set by the hook)
language = "${env:DBGPT_LANG:-zh}"
api_keys = []
encrypt_key = "your_secret_key"
# Server Configurations
[service.web]
host = "0.0.0.0"
port = 5670
[service.web.database]
type = "sqlite"
path = "pilot/meta_data/dbgpt.db"
[service.model.worker]
host = "127.0.0.1"
[rag.storage]
[rag.storage.vector]
type = "chroma"
persist_path = "pilot/data"
# Model Configurations
[models]
[[models.llms]]
name = "deepseek-reasoner"
# name = "deepseek-chat"
provider = "proxy/deepseek"
api_key = "your_deepseek_api_key"

[[serves]]
type = "file"
# Default backend for file server
default_backend = "s3"


[[serves.backends]]
type = "oss"
endpoint = "https://oss-cn-beijing.aliyuncs.com"
region = "oss-cn-beijing"
access_key_id = "${env:OSS_ACCESS_KEY_ID}"
access_key_secret = "${env:OSS_ACCESS_KEY_SECRET}"
fixed_bucket = "{your_bucket_name}"


[[serves.backends]]
# Use Tencent COS s3 compatible API as the file server
type = "s3"
endpoint = "https://cos.ap-beijing.myqcloud.com"
region = "ap-beijing"
access_key_id = "${env:COS_SECRETID}"
access_key_secret = "${env:COS_SECRETKEY}"
fixed_bucket = "{your_bucket_name}

# Model Configurations
[models]
[[models.llms]]
name = "DeepSeek-R1-Distill-Qwen-1.5B"
provider = "llama.cpp.server"
# If not provided, the model will be downloaded from the Hugging Face model hub
# uncomment the following line to specify the model path in the local file system
# https://huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-1.5B-GGUF
# path = "the-model-path-in-the-local-file-system"
path = "models/DeepSeek-R1-Distill-Qwen-1.5B-Q4_K_M.gguf

[app]
# 统一配置所有场景的温度
temperature = 0.6


[[app.configs]]
name = "chat_excel"
# 使用自定义的温度配置
temperature = 0.1
duckdb_extensions_dir = []
force_install = true


[[app.configs]]
name = "chat_normal"
memory = {type="token", max_token_limit=20000}


[[app.configs]]
name = "chat_with_db_qa"
schema_retrieve_top_k = 50
memory = {type="window", keep_start_rounds=0, keep_end_rounds=10}

docker run -it --rm -e SILICONFLOW_API_KEY=${SILICONFLOW_API_KEY} \
 -p 5670:5670 --name dbgpt eosphorosai/dbgpt-openai

bash docker/base/build_image.sh --install-mode openai

from openai import OpenAI


DBGPT_API_KEY = "dbgpt"


client = OpenAI(
    api_key=DBGPT_API_KEY,
    base_url="http://localhost:5670/api/v2",
)


messages = [
    {
        "role": "user",
        "content": "Hello, how are you?",
    },
]


has_thinking = False
reasoning_content = ""
for chunk in client.chat.completions.create(
    model="deepseek-chat",
    messages=messages,
    extra_body={
        "chat_mode": "chat_normal",
    },
    stream=True,
    max_tokens=4096,
):
    delta_content = chunk.choices[0].delta.content
    if hasattr(chunk.choices[0].delta, "reasoning_content"):
        reasoning_content = chunk.choices[0].delta.reasoning_content
    if reasoning_content:
        if not has_thinking:
            print("<thinking>", flush=True)
        print(reasoning_content, end="", flush=True)
        has_thinking = True
    if delta_content:
        if has_thinking:
            print("</thinking>", flush=True)
        print(delta_content, end="", flush=True)
        has_thinking = False

mysql -h127.0.0.1 -uroot -p{your_password} < ./assets/schema/upgrade/v0_7_0/upgrade_to_v0.7.0.sql