skills.rest
kwojtasinski-repo
kwojtasinski-repoCOMMUNITY
Explore all skills by this author →
GitHub

setup-rag-new-project

Bootstraps a Qdrant-based RAG pipeline with Python and .NET HTTP servers for new projects.

Updated Nov 19, 2020
Search Tags:#embeddings#rag#mcp#vector search#docker-compose#qdrant#document ingestion
One-click install
npx skills add https://github.com/kwojtasinski-repo/ECommerceApp --skill setup-rag-new-project-kwojtasinski-repo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: setup-rag-new-project
Source: https://github.com/kwojtasinski-repo/ECommerceApp/tree/main/.github/skills/setup-rag-new-project
Command: npx skills add https://github.com/kwojtasinski-repo/ECommerceApp --skill setup-rag-new-project-kwojtasinski-repo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Setting up retrieval-augmented generation from scratch requires wiring together a vector database, embedding models, ingestion scripts, metadata rules, and MCP-facing HTTP servers. This Skill walks through the entire bootstrap for a brand-new project so documents under docs/ and .github/context/ become embedded in Qdrant and queryable via query_docs and read_docs. ## Core Features & Use Cases - Template-driven setup: Copies canonical rag-config.yaml, metadata-rules.yaml, multilingual-glossary.yaml, and ingest/server scripts from a reference implementation. - Dual-server deployment: Provides a docker-compose stanza running Qdrant plus Python and .NET RAG HTTP servers on separate ports and collections. - Configuration guidance: Covers embedder model selection (default MiniLM 384-dim), metadata-rules auditing per folder layout, and glossary scoping by project language. - Use Case: A team splitting a service out of a monorepo uses this Skill to stand up its own RAG stack, ingest 300+ documentation chunks, and smoke-test queries through MCP within one session. ## Quick Start Ask the assistant to set up RAG for a new project named AcmeApp with the default MiniLM embedder and both Python and .NET servers.

Frequently Asked Questions about setup-rag-new-project

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I set up RAG with Qdrant for a new project?▼

Copy the canonical rag-config.yaml, metadata-rules.yaml, glossary, and ingest scripts into tools/rag, add the Qdrant and HTTP server services to docker-compose.yaml, then run the ingest script and smoke-test with a query_docs call over MCP.

Which sentence-transformers embedding model should I use for document search?▼

The default is all-MiniLM-L6-v2 at 384 dimensions, which is fast and works well for most docs. Alternatives include all-mpnet-base-v2 at 768 dimensions for higher MRR, or e5-large-v2 at 1024 dimensions for multilingual content.

Can Python and .NET RAG servers share one Qdrant collection?▼

No. Python and .NET use different chunking and metadata serialization, so mixed writes cause duplicate or stale chunks. Use separate collections such as project_docs and project_docs_dotnet, or run only one server.

Why do RAG queries return generic chunks instead of specific documents?▼

The usual cause is metadata-rules.yaml not covering your folder layout, so chunks get tagged doc_kind=other and receive no topic boost. Audit your docs directories and add a rule with an appropriate doc_kind for each uncovered path.

When should I not run a full RAG setup?▼

Skip setup if the project already has RAG and you only need to re-ingest changed content, or if the project uses context-mode only without vector search. Re-ingestion is just running the existing ingest script.

Related Skills

Explore More Software Engineering Skills→
Avila-TekAvila-Tek

documentation-and-adrs

Record architectural decisions and API changes in structured ADR templates.

Official
Intermediate
ecoDriverltdecoDriverltd

mcp-csharp-debug

Run, debug, and test C# MCP servers locally with IDE and Inspector tooling.

Official
Intermediate
randoneeringrandoneering

optimize-cortex-search-service

Optimize Cortex Search scoring weights using Bayesian optimization and ndcg evaluation.

Community
Advanced
AkanoaAkanoa
1

Moerae - AI Memory

Store and retrieve agent-authored memories via semantic search across conversation and project scopes.

Community
Advanced