--- name: llamaindex-d178df0c56eb description: >- LlamaIndex is an open-source Python framework for connecting LLMs to your own data: it loads and chunks documents, indexes them in a vector store, and answers questions over them with retrieval (RAG) and tool-calling agents. Use when ingesting documents, configuring retrieval strategies, building query engines, creating multi-step agents, or evaluating answer quality. Trigger words: llamaindex, llama-index, rag, retrieval augmented generation, vector index, query engine, document loader, knowledge base. license: Apache-2.0 compatibility: "Python 3.10+; an LLM and embedding provider (OpenAI by default, or local models through Ollama)" metadata: author: terminal-skills version: "1.1.0" category: data-ai tags: "llamaindex, rag, llm, retrieval, knowledge-base" repository: https://github.com/run-llama/llama_index --- # LlamaIndex ## Overview LlamaIndex is a data framework for building RAG pipelines, knowledge assistants, and data-augmented LLM applications. It provides document loaders, chunking strategies, several index types, hybrid retrieval with reranking, tool-calling agents, and evaluation tools for question-answering systems. The Python library is `llama-index-core` plus one pip package per integration (300+). This skill follows the 0.14 API: agents are workflow-based (`FunctionAgent`), and the agent classes from before 0.13 no longer exist. ## Instructions ### Install and configure models ```bash pip install llama-index # llama-index-core + the OpenAI LLM and embedding integrations pip install llama-index-readers-file # PDF, DOCX, PPTX, CSV parsing for SimpleDirectoryReader pip install llama-index-vector-stores-chroma llama-index-retrievers-bm25 llama-index-postprocessor-cohere-rerank ``` The package name spells the import: `llama-index-llms-ollama` is `from llama_index.llms.ollama import Ollama`. Set models once on `Settings`; every index, query engine, and evaluator picks them up. ```python from llama_index.core import Settings from llama_index.embeddings.openai import OpenAIEmbedding from llama_index.llms.openai import OpenAI # Both read OPENAI_API_KEY from the environment. Left unset, the defaults are the # older gpt-3.5-turbo and text-embedding-ada-002. Settings.llm = OpenAI(model="gpt-4o-mini") Settings.embed_model = OpenAIEmbedding(model="text-embedding-3-small") # Local models instead: pip install llama-index-llms-ollama llama-index-embeddings-ollama # from llama_index.llms.ollama import Ollama # from llama_index.embeddings.ollama import OllamaEmbedding # Settings.llm = Ollama(model="llama3.1", request_timeout=360.0, context_window=8000) # Settings.embed_model = OllamaEmbedding(model_name="nomic-embed-text") ``` ### Load, index, query ```python from llama_index.core import SimpleDirectoryReader, StorageContext, VectorStoreIndex, load_index_from_storage documents = SimpleDirectoryReader( "docs/handbook", recursive=True, required_exts=[".md", ".pdf"], filename_as_id=True ).load_data() index = VectorStoreIndex.from_documents(documents) # chunks, embeds, stores in memory index.storage_context.persist(persist_dir="storage") # Later runs: reload instead of re-embedding index = load_index_from_storage(StorageContext.from_defaults(persist_dir="storage")) response = index.as_query_engine(similarity_top_k=4).query("How many vacation days do new hires get?") print(response) for hit in response.source_nodes: print(hit.score, hit.metadata["file_name"]) ``` - When chunking, start with `SentenceSplitter` at 1024 tokens with 200 token overlap (its defaults), use `MarkdownNodeParser` for structured documents, `CodeSplitter` for code (needs `tree_sitter_language_pack`), and adjust based on evaluation results. - When indexing, use `VectorStoreIndex` as the default for most RAG, `PropertyGraphIndex` for entity relationships (`KnowledgeGraphIndex` is deprecated), and `DocumentSummaryIndex` for per-document summaries. - When building query engines, use `RetrieverQueryEngine` for standard RAG, `CitationQueryEngine` for responses with source attribution, and `SubQuestionQueryEngine` for complex multi-part queries. - Set `similarity_top_k` based on context window: 3-5 chunks for large models, 2-3 for smaller models. ### Incremental ingestion into a vector store ```python import os import chromadb from llama_index.core.extractors import TitleExtractor from llama_index.core.ingestion import IngestionPipeline from llama_index.core.node_parser import SentenceSplitter from llama_index.core.storage.docstore import SimpleDocumentStore from llama_index.vector_stores.chroma import ChromaVectorStore collection = chromadb.PersistentClient(path="chroma_db").get_or_create_collection("handbook") vector_store = ChromaVectorStore(chroma_collection=collection) pipeline = IngestionPipeline( transformations=[ SentenceSplitter(chunk_size=1024, chunk_overlap=200), TitleExtractor(), # per document: one LLM call per chunk (first 5) plus one to merge; adds document_title Settings.embed_model, # must be a stage when a vector store is attached ], docstore=SimpleDocumentStore(), # remembers a hash per document id vector_store=vector_store, ) if os.path.exists("pipeline_storage"): pipeline.load("pipeline_storage") nodes = pipeline.run(documents=documents) # unchanged files are skipped, edited ones are replaced pipeline.persist("pipeline_storage") index = VectorStoreIndex.from_vector_store(vector_store) ``` Deduplication keys on the document id, so load with `filename_as_id=True`; otherwise every run generates new ids and re-embeds everything. ### Hybrid retrieval and reranking ```python from llama_index.core.query_engine import RetrieverQueryEngine from llama_index.core.retrievers import QueryFusionRetriever from llama_index.postprocessor.cohere_rerank import CohereRerank from llama_index.retrievers.bm25 import BM25Retriever chunks = vector_store.get_nodes(node_ids=None) # every chunk in the Chroma collection hybrid = QueryFusionRetriever( [index.as_retriever(similarity_top_k=10), BM25Retriever.from_defaults(nodes=chunks, similarity_top_k=10)], similarity_top_k=10, num_queries=1, # 1 = no LLM-generated query variations mode="reciprocal_rerank", ) engine = RetrieverQueryEngine.from_args( hybrid, node_postprocessors=[CohereRerank(top_n=4)], # reads COHERE_API_KEY ) ``` ### Agents ```python import asyncio from llama_index.core.agent.workflow import FunctionAgent from llama_index.core.tools import QueryEngineTool from llama_index.core.workflow import Context handbook_tool = QueryEngineTool.from_defaults( query_engine=engine, name="handbook_search", description="Answers questions about the employee handbook: time off, expenses, hardware.", ) agent = FunctionAgent( tools=[handbook_tool], llm=Settings.llm, system_prompt="Answer HR questions. Search the handbook before answering.", ) async def main(): ctx = Context(agent) # carries chat history between runs print(await agent.run("How many vacation days do new hires get?", ctx=ctx)) print(await agent.run("And how many of them carry over?", ctx=ctx)) asyncio.run(main()) ``` `FunctionAgent` relies on the provider's native tool calling. For models without it, use `ReActAgent` from the same module; `AgentWorkflow(agents=[...])` lets several agents hand off to each other. Plain Python functions with type hints and a docstring are valid tools. ### Evaluation ```python from llama_index.core.evaluation import BatchEvalRunner, FaithfulnessEvaluator, RelevancyEvaluator async def evaluate(engine, questions): judge = OpenAI(model="gpt-4o") runner = BatchEvalRunner( {"faithfulness": FaithfulnessEvaluator(llm=judge), "relevancy": RelevancyEvaluator(llm=judge)}, workers=4, ) results = await runner.aevaluate_queries(engine, queries=questions) for name, evals in results.items(): print(name, sum(e.passing for e in evals) / len(evals)) ``` Faithfulness checks the answer against the retrieved context and relevancy against the question; neither needs labels. `CorrectnessEvaluator` scores 1-5 against a `reference` answer you supply. ## Examples ### Example 1: Build a RAG pipeline over company documentation **User request:** "Create a question-answering system over our internal docs" With `Settings`, `documents`, and the ingestion pipeline above in `ask_handbook.py`, add a citation engine: ```python import sys from llama_index.core.query_engine import CitationQueryEngine engine = CitationQueryEngine.from_args(index, similarity_top_k=4, citation_chunk_size=512) response = engine.query(sys.argv[1]) print(response) for i, hit in enumerate(response.source_nodes, start=1): print(f"[{i}] {hit.metadata['file_name']}") ``` ```bash python ask_handbook.py "How many vacation days do new hires get, and do unused days carry over?" ``` **Output** (wording varies by model): ``` New hires receive 25 vacation days per calendar year, prorated from the start date [1]. Up to 5 unused days carry over and expire on March 31 [1]. [1] time-off.md [2] expenses.md ``` The second run of the script embeds no documents (only the question), because the pipeline's document store finds every file unchanged. ### Example 2: Create a multi-source research agent **User request:** "Build an agent that can search our handbook and our runbooks, and also do date math" ```python from datetime import date from llama_index.core.agent.workflow import FunctionAgent, ToolCallResult from llama_index.core.tools import QueryEngineTool def days_between(start: str, end: str) -> int: """Number of days between two ISO dates (YYYY-MM-DD).""" return (date.fromisoformat(end) - date.fromisoformat(start)).days tools = [ QueryEngineTool.from_defaults(handbook_index.as_query_engine(), name="handbook_search", description="HR policies: time off, expenses, hardware purchases."), QueryEngineTool.from_defaults(runbook_index.as_query_engine(), name="runbook_search", description="Engineering runbooks: deploys, incident response, on-call."), days_between, ] agent = FunctionAgent(tools=tools, llm=Settings.llm, system_prompt="Use the tools; cite which one you used.") async def ask(question: str): handler = agent.run(question) async for event in handler.stream_events(): if isinstance(event, ToolCallResult): print("tool:", event.tool_name, event.tool_kwargs) print(await handler) ``` **Output** for `asyncio.run(ask("Vacation carry-over expires March 31. How many days is that from 2026-10-01?"))` (which tools are called, and with what arguments, varies by model): one `tool:` line per call — for example `handbook_search`, then `days_between {'start': '2026-10-01', 'end': '2027-03-31'}` — followed by an answer that uses the returned 181 days. ## Guidelines - Use `SentenceSplitter` with 1024 token chunks and 200 token overlap as the starting point. - Install `llama-index-readers-file` before loading PDFs or Office files: since 0.14.18 `pip install llama-index` no longer pulls it in, and without it `SimpleDirectoryReader` only logs a warning and reads those files as raw text. - Code written for 0.12 or earlier breaks: `ServiceContext` raises (use `Settings`), and `ReActAgent.from_tools`, `AgentRunner`, `FunctionCallingAgent`, `OpenAIAgent`, and `QueryPipeline` were removed in 0.13. Agents are async; call them with `await agent.run(...)`. - Use hybrid retrieval (vector + keyword) for production; pure vector search misses exact term matches. `QueryFusionRetriever` runs its retrievers asynchronously, so a vector store created with only a sync client (Qdrant, for one) fails there — pass its async client too or set `use_async=False`. - Add a reranker (`CohereRerank`) after retrieval to improve result relevance for small cost: retrieve 10, keep 4. - Metadata extractors (`TitleExtractor`, `SummaryExtractor`) improve retrieval but cost LLM calls on every new or changed document; add them once the plain pipeline is measured. - Evaluate with `BatchEvalRunner` on a fixed question set before deploying and after every change to chunking or retrieval; subjective quality assessment does not scale. - Use `IngestionPipeline` with a docstore for incremental updates; do not re-embed unchanged documents. Persist both the pipeline state and the vector store. - Documents and questions are sent to the configured LLM and embedding provider. For data that must stay on the machine, set local models on `Settings` before building any index, and re-embed after changing the embedding model — vectors from different models are not comparable. - The maintainers now focus on the hosted LlamaParse platform and keep the framework as an open toolkit. For hard PDFs (scans, tables) the built-in readers are basic; for a single prompt over a small file, call the model API directly instead of building an index. --- Source attribution: terminal-skills Original project: https://github.com/TerminalSkills/skills/blob/8a9852d489f274538be9bf1c9121aeb403750b1c/skills/llamaindex/SKILL.md License: Apache-2.0 License information: https://www.apache.org/licenses/LICENSE-2.0 Original upstream instructions; downloaded from Mega AI Hub. ## Upstream license notice Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for use, reproduction, and distribution as defined by Sections 1 through 9 of this document. "Licensor" shall mean the copyright owner or entity authorized by the copyright owner that is granting the License. "Legal Entity" shall mean the union of the acting entity and all other entities that control, are controlled by, or are under common control with that entity. For the purposes of this definition, "control" means (i) the power, direct or indirect, to cause the direction or management of such entity, whether by contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the outstanding shares, or (iii) beneficial ownership of such entity. "You" (or "Your") shall mean an individual or Legal Entity exercising permissions granted by this License. "Source" form shall mean the preferred form for making modifications, including but not limited to software source code, documentation source, and configuration files. "Object" form shall mean any form resulting from mechanical transformation or translation of a Source form, including but not limited to compiled object code, generated documentation, and conversions to other media types. "Work" shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work (an example is provided in the Appendix below). "Derivative Works" shall mean any work, whether in Source or Object form, that is based on (or derived from) the Work and for which the editorial revisions, annotations, elaborations, or other modifications represent, as a whole, an original work of authorship. For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof. "Contribution" shall mean any work of authorship, including the original version of the Work and any modifications or additions to that Work or Derivative Works thereof, that is intentionally submitted to the Licensor for inclusion in the Work by the copyright owner or by an individual or Legal Entity authorized to submit on behalf of the copyright owner. For the purposes of this definition, "submitted" means any form of electronic, verbal, or written communication sent to the Licensor or its representatives, including but not limited to communication on electronic mailing lists, source code control systems, and issue tracking systems that are managed by, or on behalf of, the Licensor for the purpose of discussing and improving the Work, but excluding communication that is conspicuously marked or otherwise designated in writing by the copyright owner as "Not a Contribution." "Contributor" shall mean Licensor and any individual or Legal Entity on behalf of whom a Contribution has been received by the Licensor and subsequently incorporated within the Work. 2. Grant of Copyright License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable copyright license to reproduce, prepare Derivative Works of, publicly display, publicly perform, sublicense, and distribute the Work and such Derivative Works in Source or Object form. 3. Grant of Patent License. Subject to the terms and conditions of this License, each Contributor hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work, where such license applies only to those patent claims licensable by such Contributor that are necessarily infringed by their Contribution(s) alone or by combination of their Contribution(s) with the Work to which such Contribution(s) was submitted. If You institute patent litigation against any entity (including a cross-claim or counterclaim in a lawsuit) alleging that the Work or a Contribution incorporated within the Work constitutes direct or contributory patent infringement, then any patent licenses granted to You under this License for that Work shall terminate as of the date such litigation is filed. 4. Redistribution. You may reproduce and distribute copies of the Work or Derivative Works thereof in any medium, with or without modifications, and in Source or Object form, provided that You meet the following conditions: (a) You must give any other recipients of the Work or Derivative Works a copy of this License; and (b) You must cause any modified files to carry prominent notices stating that You changed the files; and (c) You must retain, in the Source form of any Derivative Works that You distribute, all copyright, patent, trademark, and attribution notices from the Source form of the Work, excluding those notices that do not pertain to any part of the Derivative Works; and (d) If the Work includes a "NOTICE" text file as part of its distribution, then any Derivative Works that You distribute must include a readable copy of the attribution notices contained within such NOTICE file, excluding any notices that do not pertain to any part of the Derivative Works, in at least one of the following places: within a NOTICE text file distributed as part of the Derivative Works; within the Source form or documentation, if provided along with the Derivative Works; or, within a display generated by the Derivative Works, if and wherever such third-party notices normally appear. The contents of the NOTICE file are for informational purposes only and do not modify the License. You may add Your own attribution notices within Derivative Works that You distribute, alongside or as an addendum to the NOTICE text from the Work, provided that such additional attribution notices cannot be construed as modifying the License. You may add Your own copyright statement to Your modifications and may provide additional or different license terms and conditions for use, reproduction, or distribution of Your modifications, or for any such Derivative Works as a whole, provided Your use, reproduction, and distribution of the Work otherwise complies with the conditions stated in this License. 5. Submission of Contributions. Unless You explicitly state otherwise, any Contribution intentionally submitted for inclusion in the Work by You to the Licensor shall be under the terms and conditions of this License, without any additional terms or conditions. Notwithstanding the above, nothing herein shall supersede or modify the terms of any separate license agreement you may have executed with Licensor regarding such Contributions. 6. Trademarks. This License does not grant permission to use the trade names, trademarks, service marks, or product names of the Licensor, except as required for reasonable and customary use in describing the origin of the Work and reproducing the content of the NOTICE file. 7. Disclaimer of Warranty. Unless required by applicable law or agreed to in writing, Licensor provides the Work (and each Contributor provides its Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied, including, without limitation, any warranties or conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are solely responsible for determining the appropriateness of using or redistributing the Work and assume any risks associated with Your exercise of permissions under this License. 8. Limitation of Liability. In no event and under no legal theory, whether in tort (including negligence), contract, or otherwise, unless required by applicable law (such as deliberate and grossly negligent acts) or agreed to in writing, shall any Contributor be liable to You for damages, including any direct, indirect, special, incidental, or consequential damages of any character arising as a result of this License or out of the use or inability to use the Work (including but not limited to damages for loss of goodwill, work stoppage, computer failure or malfunction, or any and all other commercial damages or losses), even if such Contributor has been advised of the possibility of such damages. 9. Accepting Warranty or Additional Liability. While redistributing the Work or Derivative Works thereof, You may choose to offer, and charge a fee for, acceptance of support, warranty, indemnity, or other liability obligations and/or rights consistent with this License. However, in accepting such obligations, You may act only on Your own behalf and on Your sole responsibility, not on behalf of any other Contributor, and only if You agree to indemnify, defend, and hold each Contributor harmless for any liability incurred by, or claims asserted against, such Contributor by reason of your accepting any such warranty or additional liability. END OF TERMS AND CONDITIONS Copyright 2025 Terminal Skills Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. # Mega AI Hub resource metadata Creator / publisher: terminal-skills Original project: https://github.com/TerminalSkills/skills/blob/8a9852d489f274538be9bf1c9121aeb403750b1c/skills/llamaindex/SKILL.md Attribution basis: Source repository record Fields of work: Data & Analytics, Development & Engineering, Product Management Roles: Data Scientist, ML / AI Engineer, Product Manager, Growth Product Manager, Technical Product Manager Specialties: API, Backend Tools: Llamaindex Country applicability: Country not specified Classification basis: Source category, role and specialization labels Instruction SHA-256: 768d1d6e7169ede5f79755b331aec09be939a2ead0312dc68d14464b671ce012