一键在 Manus 中运行任何 Skill

开始使用

mongodb

星标2

分支1

更新时间2026年1月15日 19:24

Expert guidance on MongoDB implementation for RAG, including aggregation pipelines and search patterns.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

FullFran

FullFran/Hybrid-RAG-example

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

MongoDB Expert Skill

This skill provides patterns for implementing RAG logic with MongoDB Atlas.

📂 Storage Pattern

Two-Collection Pattern:
- documents: Stores source document text and global metadata.
- chunks: Stores text fragments, embeddings (list[float]), and a foreign key (document_id) to the source document.
Score Meta: Use {"$meta": "vectorSearchScore"} for semantic scores and {"$meta": "searchScore"} for text scores.

🔍 Search Patterns

Semantic Search: Use $vectorSearch aggregation. Default numCandidates: 100.
Text Search: Use $search (Atlas Search) in chunks collection.
Hybrid Search: Currently implemented via manual Reciprocal Rank Fusion (RRF) in RAGService by merging results from semantic and text search.
Joins: Always use $lookup to fetch document metadata (title, source) when returning search results.

🛠️ Code Standards

Async First: Use motor (AsyncIOMotorClient) for all database operations.
List Embeddings: Embeddings MUST be stored and queried as Python lists of floats (e.g., 1536 dims for OpenAI).
ID Handling: Convert string IDs to bson.ObjectId where necessary.
Graceful Failure: Handle missing search indexes (code 291) with clear error messages.

同仓库更多 Skills

同仓库

architecture-decision-record

FullFran/Hybrid-RAG-example

Help the team document and maintain Architecture Decision Records (ADRs).

2026-02-022

docling

FullFran/Hybrid-RAG-example

Expert guidance on document processing with Docling and audio transcription with Whisper.

2026-02-022

docs-standard

FullFran/Hybrid-RAG-example

Standard for creating technical documentation in this repository. Use this when writing new documentation in docs/ to ensure consistent hierarchy and formatting.

2026-02-022

mermaid-diagrams

FullFran/Hybrid-RAG-example

Expert guidance on creating accurate, visually polished Mermaid diagrams for architecture documentation.

2026-02-022

mongodb

FullFran/Hybrid-RAG-example

Expert guidance on MongoDB implementation for RAG, including aggregation pipelines and search patterns.

2026-02-022

pydantic-ai

FullFran/Hybrid-RAG-example

Expert guidance on building agents and tools with Pydantic AI.

2026-02-022

name	mongodb
description	Expert guidance on MongoDB implementation for RAG, including aggregation pipelines and search patterns.

MongoDB Expert Skill

This skill provides patterns for implementing RAG logic with MongoDB Atlas.

📂 Storage Pattern

Two-Collection Pattern:
- documents: Stores source document text and global metadata.
- chunks: Stores text fragments, embeddings (list[float]), and a foreign key (document_id) to the source document.
Score Meta: Use {"$meta": "vectorSearchScore"} for semantic scores and {"$meta": "searchScore"} for text scores.

🔍 Search Patterns

Semantic Search: Use $vectorSearch aggregation. Default numCandidates: 100.
Text Search: Use $search (Atlas Search) in chunks collection.
Hybrid Search: Currently implemented via manual Reciprocal Rank Fusion (RRF) in RAGService by merging results from semantic and text search.
Joins: Always use $lookup to fetch document metadata (title, source) when returning search results.

🛠️ Code Standards

Async First: Use motor (AsyncIOMotorClient) for all database operations.
List Embeddings: Embeddings MUST be stored and queried as Python lists of floats (e.g., 1536 dims for OpenAI).
ID Handling: Convert string IDs to bson.ObjectId where necessary.
Graceful Failure: Handle missing search indexes (code 291) with clear error messages.