Exécutez n'importe quel Skill dans Manus
en un clic

Exécutez n'importe quel Skill dans Manus en un clic

Commencer

mongodb

Étoiles2

Forks1

Mis à jour15 janvier 2026 à 19:24

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

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

FullFran

FullFran/Hybrid-RAG-example

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Architectes de bases de donnéesProfessions informatiques et mathématiques·SOC 15-1243

SKILL.md

readonly

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.

Plus depuis ce dépôt

même dépôt

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.