Skip to main content
Glama
msstnk

qdrant-mcp

by msstnk

qdrant-mcp

用于 Qdrant 文档摄入和语义搜索的 MCP 服务器。

概述

qdrant-mcp 提供的工具可以:

  • 将本地文档摄入到 Qdrant 集合中

  • 使用 OpenAI 生成嵌入

  • 运行带有可选元数据过滤的向量搜索

Related MCP server: mcp-server-qdrant

功能

  • ingest_documents

    • 通过 MarkItDown 将 docxpptxpdf 等文件转换为 Markdown

    • 使用 chunk_sizeoverlap_ratio 将内容拆分为块

    • 使用 OpenAI Embeddings(默认为 text-embedding-3-small)对块进行嵌入

    • 将块文本和元数据插入/更新到 Qdrant

  • search_documents

    • 使用相同的嵌入 API 对查询文本进行嵌入

    • 从 Qdrant 检索前 k 个匹配项

    • 支持按 categorypath 进行过滤

要求

  • Python 3.11+

  • uv

  • Qdrant(例如 http://localhost:6333

  • OPENAI_API_KEY

设置

uv sync

在 Codex CLI 中运行

[mcp_servers.qdrant-mcp]
command = "uv"
args = ["run", "qdrant-mcp"]
cwd = "/sandbox/qdrant-mcp"

env = {
  OPENAI_API_KEY = "sk-...",
  QDRANT_URL = "http://127.0.0.1:6333",
  QDRANT_API_KEY = "QDRANT_API_KEY",
  QDRANT_COLLECTION = "codex_collection",
  CHUNK_HEADER_MODEL = "gpt-5.4-mini"
}

测试

.env 中设置 OPENAI_API_KEYQDRANT_URLQDRANT_API_KEY,然后运行:

uv run python -m unittest tests/integration/test_qdrant_integration.py

MCP 工具

ingest_documents

参数:

  • paths: list[str]

  • category: str

  • chunk_size: int = 1200

  • overlap_ratio: float = 0.15

  • embedding_model: str = "text-embedding-3-small"

  • chunk_header_mode: Literal["enabled", "disabled"] = "enabled"

返回:

  • collection

  • embedding_model

  • ingested_files

  • ingested_points

  • failed_files

search_documents

参数:

  • query: str

  • top_k: int = 5

  • category: str | None = None

  • path: str | None = None

  • embedding_model: str = "text-embedding-3-small"

返回:

  • collection

  • embedding_model

  • query

  • count

  • results (score, path, category, chunk_index, text)

delete_documents_by_path

参数:

  • path: str

  • category: str | None = None

返回:

  • collection

  • path

  • category

  • status

  • operation_id

list_category

参数:

  • limit: int = 100

返回:

  • collection

  • count

  • categories

list_path

参数:

  • category: str

  • limit: int = 1000

返回:

  • collection

  • category

  • count

  • paths

注意事项

  • 如果目标集合不存在,它会在首次摄入时自动创建。

  • 如果 categorypath 的负载索引不存在,它们会在摄入过程中创建。

  • 默认情况下,摄入过程会在每个块前添加一个生成的 Chunk-Header(最大 64 字符),该头部派生自前 4096 字节。

  • chunk_header_modeenabled(默认值:gpt-5.4-mini)时,Chunk-Header 模型从 CHUNK_HEADER_MODEL 读取。

  • 集合名称仅通过 QDRANT_COLLECTION 配置(而非通过 MCP 工具参数)。

  • 使用 text-embedding-3-small 时,向量大小为 1536

许可证

参见 LICENSE

A
license - permissive license
-
quality - not tested
D
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

  • A
    license
    A
    quality
    D
    maintenance
    Local-first RAG indexing and semantic search MCP server. Enables document retrieval and context-aware queries using local embedding models.
    3
    9
    MIT
  • A
    license
    -
    quality
    C
    maintenance
    MCP server for Qdrant vector database with local BERT embeddings. Enables semantic search and vector storage operations through natural language.
    MIT
  • A
    license
    -
    quality
    D
    maintenance
    MCP server that searches documents in Qdrant using embeddings from LMStudio. Takes a text query, converts it to a vector via LMStudio's OpenAI-compatible API, and performs semantic search in Qdrant.
    13
    ISC

View all related MCP servers

Related MCP Connectors

View all MCP Connectors

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/msstnk/qdrant-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server