context-llm-pipeline
RAG pipeline, embeddings, LLM interactions, and flow orchestration.
When & Why to Use This Skill
The context-llm-pipeline skill is a comprehensive framework designed for managing RAG (Retrieval-Augmented Generation) workflows, embeddings, and LLM context orchestration. It provides a structured approach to document ingestion, vector retrieval, and complex AI flow management, enabling developers to build highly context-aware AI agents. By centralizing LLM service orchestration and search pipeline logic, it ensures efficient prompt engineering and accurate information retrieval for sophisticated AI applications.
Use Cases
- Developing and optimizing RAG pipelines to provide AI agents with access to proprietary or real-time knowledge bases.
- Automating document ingestion and processing flows to keep vector databases synchronized for accurate retrieval.
- Orchestrating multi-step LLM workflows that involve web scraping, content analysis, and automated template reviews.
- Implementing centralized LLM service handlers to manage complex interactions between different models and research integrations like ZAI.
- Enhancing AI agent memory by designing efficient search pipelines and vector search logic for better context management.
| name | context-llm-pipeline |
|---|---|
| description | RAG pipeline, embeddings, LLM interactions, and flow orchestration. |
| tags | [backend, ai, rag] |
LLM Pipeline Context
Overview
Core AI logic including RAG flows, LLM service orchestration, and vector retrieval.
Active Files
Orchestration (Flows)
backend/flows/ingestion_flow.py- Document ingestionbackend/flows/scraping_flow.py- Scrape orchestrationbackend/flows/template_review_flow.py- LLM review flow
Services
backend/services/llm/orchestrator.py- Main LLM handlerbackend/services/llm/pipeline.py- Pipeline logicbackend/services/llm/analyzer.py- Analysis logicbackend/services/research/zai.py- ZAI research integrationbackend/services/search_pipeline_service.py- Search pipeline
Shared Packages
packages/llm-common/- Shared types and utilities (Submodule)
Usage
Use this skill when working on RAG, prompt engineering, or vector search logic.