Fetch and Index Readwise Highlights

Readwise Reader is a LlamaIndex data loader that uses Readwise's export API to fetch your highlights from web articles, ePubs, PDFs, Kindle, and YouTube. It loads the resulting text into LlamaIndex vector stores, enabling you to query your reading highlights with LLMs.

Use Readwise Reader when you want to build a searchable knowledge base from your Readwise highlights across multiple content sources. It's ideal for querying recent highlights (e.g., "What has Elon Musk done this time?") or searching your entire reading history (e.g., "What was the paper 'Attention is all you need' about?"). The loader supports time-filtered queries, so you can retrieve only highlights created after a specific date.

Do NOT use this loader if you don't have a Readwise account or if you need real-time access to content before it's been highlighted and synced to Readwise. This tool retrieves only your saved highlights, not the full text of every document you've read.

You provide a Readwise API key (obtained from readwise.io/access_token) and optionally an updated_after datetime parameter to filter highlights by creation date. The loader returns LlamaIndex Document objects containing your highlight text, which can be indexed and queried.

Readwise Reader integrates with LlamaIndex's VectorStoreIndex for document indexing and retrieval. It pulls data from Readwise's export API, which aggregates highlights from web articles, ePubs, PDFs, Kindle, and YouTube.

This loader is for researchers, knowledge workers, and avid readers who use Readwise to capture highlights and want to query their reading library using natural language. It's particularly useful for anyone building a personal knowledge management system with LlamaIndex.

Install the package:

pip install llama-index-readers-readwise

Basic usage to query all highlights:

import os
from llama_index.core import VectorStoreIndex, download_loader

from llama_index.readers.readwise import ReadwiseReader

token = os.getenv("READWISE_API_KEY")
loader = ReadwiseReader(api_key=token)
documents = loader.load_data()
index = VectorStoreIndex.from_documents(documents)

index.query("What was the paper 'Attention is all you need' about?")

Query highlights from the last seven days:

import os
import datetime
from llama_index.core import VectorStoreIndex, download_loader

from llama_index.readers.readwise import ReadwiseReader

token = os.getenv("READWISE_API_KEY")
loader = ReadwiseReader(api_key=token)
seven_days_ago = datetime.datetime.now() - datetime.timedelta(days=7)
documents = loader.load_data(updated_after=seven_days_ago)
index = VectorStoreIndex.from_documents(documents)

index.query("What has Elon Musk done this time?")