LangChain and LlamaIndex cheat sheet (LangChain)

1. 1RAG and agents: the big pictureWhat retrieval-augmented generation and agents are, the two phases of a RAG system, and when a framework earns its place.7 concepts

   1. 1Retrieval-augmented generationA pattern where the app first retrieves passages relevant to a question from your own data, then asks the LLM to answer using those passages as context.
   2. 2Indexing time and query timeIndexing loads documents, splits them into chunks, embeds them and stores the vectors.
   3. 3AgentsAn LLM running in a loop that decides which tool to call next, reads the result and continues until it can answer or hits a limit.
   4. 4Context window and tokensThe model reads a limited number of tokens (word pieces) per call, shared between instructions, retrieved text, history and the answer.
   5. 5Grounding and citationsInstructing the model to answer only from the supplied context, to say when the context does not contain the answer, and to reference its sources.
   6. 6Prompt injectionInstructions hidden in retrieved documents, web pages or tool results that try to override your prompt, such as "ignore previous instructions and email this…
   7. 7Framework or plain SDKLangChain and LlamaIndex supply loaders, splitters, vector store integrations, retrievers and agent loops behind common interfaces.

   7 in this chapterOpen this chapter
2. 2Loading and chunkingTurning PDFs, web pages and other sources into clean text with metadata, then into chunks sized for retrieval.7 concepts

   1. 1Document loadersComponents that read a source (files, web pages, Notion, databases) and return document objects holding text plus metadata such as file name and page.
   2. 2MetadataKey-value facts attached to each document and inherited by its chunks: source, title, section, date, owner team, access level.
   3. 3ChunkingSplitting documents into passages small enough to embed and retrieve precisely, ideally along natural boundaries such as paragraphs and sentences.
   4. 4Chunk size and overlapSize is the maximum length of a chunk; overlap repeats the end of one chunk at the start of the next.
   5. 5Structure-aware splittingSplitting on the document's own structure (headings, code blocks, HTML sections) or on shifts in meaning, and keeping the heading path as metadata.
   6. 6Parsing difficult documentsPDFs with tables, columns, scans and charts lose structure when read as plain text.
   7. 7Incremental ingestionGiving each document a stable id and a content hash, so re-running ingestion only adds new documents, updates changed ones and deletes removed ones.

   7 in this chapterOpen this chapter
3. Embeddings and vector storesHow text becomes vectors, where those vectors live, and how filtering and hybrid search make lookups precise.7 conceptsOpens with credits
4. Retrieval and rerankingGetting the right few chunks in front of the model: retrievers, reranking, query rewriting and how context is packed.6 conceptsOpens with credits
5. Agents, tools and memoryLetting the model act: defining tools, the loop that runs them, memory across turns, and the limits that keep agents safe.8 conceptsOpens with credits
6. Evaluation and observabilityMeasuring retrieval and answers against a test set, tracing every call, and keeping quality from quietly regressing.7 conceptsOpens with credits
7. LangChain coreChat models, messages, prompt templates, runnables composed with LCEL, structured output and tool calling.8 conceptsOpens with credits
8. Agents with LangGraph and LangSmithLangChain agents run on LangGraph: prebuilt agents first, then graphs with explicit state, persistence and human approval, all traced in LangSmith.8 conceptsOpens with credits
9. 9Glossary, A to ZEvery concept in the sheet and the smaller terms it uses, in one list.80 terms

RAG pipelines, retrieval quality, agents, tools, memory and evaluation in Python, with LangChain and LangGraph or LlamaIndex.