LlamaIndex is a data framework that connects LLMs to your data and enables Retrieval Augmented Generation and agent workflows.

When should I use LlamaIndex?

For RAG over documents and databases, querying structured data with natural language, and building research agents that browse and summarize information.

What is Workflows in LlamaIndex?

An event-driven architecture that replaces traditional DAGs with @step listeners to orchestrate complex, modular RAG pipelines.

llamaindex

npx machina-cli add skill G1Joshi/Agent-Skills/llamaindex --openclaw

Files (1)

SKILL.md

1.2 KB

LlamaIndex

LlamaIndex (formerly GPT Index) connects LLMs to your data. 2025 introduces Workflows, an event-driven way to build complex RAG pipelines.

When to Use

RAG (Retrieval Augmented Generation): Indexing PDFs, Docs, SQL to chat with them.
Structured Data: Querying SQL/Pandas with natural language (NLSQL).
Agents: Building research agents that browse the web and summarize.

Core Concepts

Workflows

Event-driven architecture for agents. Replace DAGs with event listeners (@step).

Query Engine

High-level API (index.as_query_engine()) to ask questions.

Data Loaders (LlamaHub)

Connectors for Notion, Slack, Discord, PDF, etc.

Best Practices (2025)

Do:

Use Workflows: They are harder to learn but easier to debug than monolithic engines.
Use Hybrid Search: BM25 (Keyword) + Vector Search for best retrieval accuracy.
Use Rerankers: Always rerank retrieved nodes (Cohere/BGE) before sending to LLM.

Don't:

Don't dump raw text: Use "Node Parsers" to chunk data intelligently (Markdown, Semantic).

References

LlamaIndex Documentation

Source

git clone https://github.com/G1Joshi/Agent-Skills/blob/main/skills/ai-ml/llamaindex/SKILL.mdView on GitHub

Overview

LlamaIndex connects LLMs to your data to enable Retrieval Augmented Generation and AI agents. The 2025 release introduces Workflows, an event-driven approach to building complex RAG pipelines.

How This Skill Works

Use index.as_query_engine() to interrogate indexed data with natural language. Data Loaders (LlamaHub) connect sources like Notion, Slack, Discord, and PDFs, while Workflows coordinates an event-driven pipeline and optional reranking before delivering answers.

When to Use It

RAG: index PDFs, docs, and SQL to chat with your data
Structured data: natural language queries over SQL or Pandas data (NLSQL)
Agents: build research agents that browse the web and summarize findings
Data integration: connect Notion, Slack, Discord, and PDFs via LlamaHub
End-to-end RAG pipelines: orchestrate data loading, retrieval, and answer generation with Workflows

Quick Start

Step 1: Identify data sources and load them with LlamaHub connectors (Notion, Slack, PDF, etc.) to create an index
Step 2: Build a Workflows-based RAG pipeline with event-driven steps (@step) for loading, retrieval, and reranking
Step 3: Use index.as_query_engine() to ask questions and apply rerankers for higher-quality results

Best Practices

Use Workflows: modular, debuggable pipelines are easier to maintain than monolithic engines
Use Hybrid Search: combine BM25 keyword search with vector search for best retrieval accuracy
Use rerankers before sending results to the LLM to improve answer quality
Don't dump raw text: use Node Parsers to chunk data intelligently (Markdown, semantic)
Keep data sources connected and up-to-date via LlamaHub connectors for reliable loading

Example Use Cases

Index legal PDFs and contracts to enable quick QA and redlining in a corporate wiki
Index product docs and chat with them to support customer inquiries
Run NLQ over SQL/Pandas dashboards to extract metrics and insights
Deploy a research agent that browses the web and summarizes sources for literature reviews
Build an onboarding knowledge base in Notion and answer employee questions

Frequently Asked Questions

Add this skill to your agents