--- title: "Readwise Integration" description: "Readwise is the highlights library and Readwise Reader is its read-later app. Agents save and tag highlights and books, pull the daily review, and create, list, update, and tag Reader documents from one token." url: https://flowrunner.ai/integrations/readwise date_modified: 2026-09-04T14:52:04-07:00 --- # Readwise [Productivity](https://flowrunner.ai/integrations/category/project-management-productivity) Readwise is the highlights library and Readwise Reader is its read-later app. Agents save and tag highlights and books, pull the daily review, and create, list, update, and tag Reader documents from one token. [Verified](https://flowrunner.ai/integrations/verified "What does verified mean?") · 25 actions · API key · available [Readwise website](https://readwise.io/) · [Platform Documentation](https://readwise.io/api_deets) · Capability data verified 2026-08-24 1. A scheduled sync runs, or a source worth reading arrives from a monitoring feed or a teammate 2. Save Document sends the URL to Readwise Reader, which fetches and parses the page itself 3. Export Highlights pulls books and articles with their highlights nested inside, using the last run's timestamp so the pull stays incremental 4. The agent asks for deleted highlights explicitly, because they are left out otherwise and a mirror that never asks never learns anything was removed 5. The agent separates real articles from the highlights and notes that Reader also returns as documents in the same list 6. Update Documents In Bulk files what has been dealt with into Archive, leaving every field it does not mention alone 7. Before a person's highlights are copied into a shared destination, that person approves what is included ## What This Integration Enables Readwise is two products behind one token. The highlights library is a stream you push into and pull out of, not a database you edit; Readwise's own guidance is that two operations cover almost everything an integration needs, with Save Highlights writing in and Export Highlights reading out, and the rest existing for careful editing rather than syncing. Readwise Reader is the second product, a read later queue where saving a URL makes Reader fetch and parse the page itself. The same token reaches both. FlowRunner agents save highlights and documents, export the library incrementally, tag highlights and sources, read and complete the daily review, and file Reader documents in bulk. Every action is built and verified against Readwise's published v2 and v3 APIs. The behavior worth designing around is that the library de-duplicates silently. Readwise matches on title, author, text, and source URL together, so re-sending an identical highlight does nothing rather than creating a second one. That makes a retry safe and makes a badly built correction pass invisible, since the flow reports success either way. Saving a highlight with a highlight URL turns that URL into an update handle, which is the supported way to change something already sent. This is a connector where the honest [human-in-the-loop](https://flowrunner.ai/concepts/human-in-the-loop) gate is not about a destructive write at all, and the page says exactly where it does belong further down. ### Without FlowRunner **Reading and writing are separate lives**: What the team reads never reaches the documents the team produces **The queue grows without triage**: Saved articles pile up because filing them is a chore nobody schedules **Deletions do not travel**: A highlight removed on purpose keeps reappearing wherever it was copied ### With FlowRunner **Sources arrive where the work happens**: Highlights land beside the brief, the doc, or the ticket that needed them **The queue triages itself**: An agent files what has been handled and surfaces what has not, on a schedule **Removals are honored**: The sync asks for deletions every run, so taking a highlight out actually takes it out ## Use Case Scenarios ### A research pipeline that ends in the brief The team keeps its reading in Readwise. When a brief needs sources, the agent calls Export Highlights with the timestamp of the last pull, which returns books and articles with their highlights nested inside them, and it asks for deleted highlights so the mirror stays honest. It filters to the tags the brief cares about with List Highlight Tags and List Book Tags, then assembles the passages with their sources into a [Google Docs](https://flowrunner.ai/integrations/google-docs) document or a page in [Notion](https://flowrunner.ai/integrations/notion). The writer starts from quoted material with attribution attached rather than from a blank page and a memory of having read something useful. ### Capture into Reader from wherever the link showed up A link lands in a [Slack](https://flowrunner.ai/integrations/slack) channel, arrives on a monitoring alert, or comes through a form in [Typeform](https://flowrunner.ai/integrations/typeform). The agent calls Save Document with the URL, and Reader fetches and parses the page itself; where the source is behind something Reader cannot reach, the agent sends the HTML it already has instead and Reader saves exactly that. It sets the tags and the location on the way in, so the item arrives triaged rather than joining an undifferentiated inbox. Saving a URL Reader already holds does not duplicate it, which means the same flow can run from three sources without any coordination between them. ### An inbox that files itself Once a week the agent calls List Documents and filters by category, which matters here because highlights and notes made inside Reader come back in that same list as documents with a parent id naming the article they belong to. A flow that expects only articles and does not filter will misread its own queue. The agent identifies what has already been quoted in published work, what has sat untouched past the age the team agreed on, and what carries a tag that marks it as reference material. Update Documents In Bulk moves the first two groups to Archive in one call, leaving every field it does not mention untouched, and a short summary of what moved posts back to the channel. Anything the agent is unsure about stays in the inbox rather than being filed on a guess. ## Human-in-Loop Highlight The risk on this connector is not that an agent writes something wrong. It is that an agent copies something private into a place it does not belong. Highlights are marginalia. They are the sentences a person underlined and the notes they wrote to themselves, and Export Highlights returns all of it for the whole account with a single token. A flow that mirrors one person's library into a shared team space is publishing their reading, including the parts they highlighted because they disagreed. So the gate on this page sits in front of the export, not in front of any write. Before the first sync into a shared destination, the agent posts to the account holder: "This will copy 1,240 highlights across 86 sources from your Readwise library into the team knowledge base, refreshed weekly. Include everything, restrict it to sources tagged `team`, or restrict it to a list you approve? Deletions will be pulled every run, so anything you remove in Readwise disappears from the shared copy too." The person sets the scope, and the agent never widens it on its own. The library is theirs; what leaves it is their decision. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 25 actions ### Library 5 - **Save Highlights** Saves highlights into the library, creating the book, article, or podcast they belong to. Readwise says this is the only endpoint an application that saves highlights needs. Highlights are de-duplicated on title, author, text, and source URL together, so re-sending an identical one does nothing. - **Export Highlights** Returns books and articles with all of their highlights nested inside. The read side of the same pair: run it with no parameters first and page through, then pass a timestamp on later runs. Deleted highlights are left out unless they are asked for. - **Check Token** Checks the access token without reading any data. A valid token answers with an empty body, so there is nothing to inspect beyond the fact that it did not fail. - **Get Daily Review** Returns today's daily review with the highlights Readwise chose to resurface, and whether it has been completed. - **Complete Daily Review** Marks today's daily review complete, the same as finishing it in the app. ### Highlights 8 - **List Highlights** Returns highlights one page at a time with filters. Rate limited well below the rest of the API, so Export Highlights is the operation built for bulk reading. - **Get Highlight** Returns one highlight with its note, color, location, and tags. - **Update Highlight** Changes a highlight's text, note, location, URL, or color. Fields left empty are untouched. - **Delete Highlight** Deletes a highlight. A mirror only learns it went if its export asks for deleted highlights. - **List Highlight Tags** Returns the tags on one highlight. - **Add Highlight Tag** Adds a tag to a highlight. Readwise's own documentation warns that inline tagging inside a note will not recreate a tag once it has been deleted, so tagging through this operation is the durable path. - **Update Highlight Tag** Renames one tag on one highlight. - **Delete Highlight Tag** Removes a tag from a highlight. ### Sources 6 - **List Books** Returns the sources a highlight can belong to: books, articles, tweets, podcasts, and supplementals. Rate limited like List Highlights. - **Get Book** Returns one source with its highlight count, cover image, and tags. - **List Book Tags** Returns the tags on one source. - **Add Book Tag** Adds a tag to a source, which tags the whole book or article rather than one highlight. The right level for marking a source as team reference material. - **Update Book Tag** Renames one tag on one source. - **Delete Book Tag** Removes a tag from a source. ### Reader 6 - **Save Document** Saves a document to Readwise Reader. Give a URL and Reader fetches and parses the page itself; give HTML and it saves exactly what was sent. Saving a URL already in Reader does not duplicate it. - **List Documents** Returns Reader documents. Highlights and notes made inside Reader are documents too, carrying a parent id naming the article they belong to, so a workflow expecting only articles has to filter by category. - **Update Document** Changes one Reader document, leaving empty fields untouched. Setting the location to Archive is how a workflow files something away once it has been dealt with. - **Update Documents In Bulk** Changes several Reader documents in one call, each entry setting different fields, and anything an entry does not mention is left alone on that document. - **Delete Document** Deletes a Reader document along with the highlights and notes made inside it. Moving it to Archive through Update Document is the reversible alternative and should be the default in any automated triage. - **List Document Tags** Returns every tag used across Reader documents, which is how a workflow reads the taxonomy before applying to it. ## Frequently Asked Questions ### What can FlowRunner do with Readwise? FlowRunner agents can run Save Highlights, Export Highlights, and Check Token in Readwise, plus 22 more actions. ### Does connecting Readwise to FlowRunner require OAuth? No. Readwise connects to FlowRunner with an API key, no OAuth flow required. ### Can Readwise trigger a FlowRunner workflow automatically? Readwise doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at Readwise?** This integration exposes Readwise to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/readwise. Site index: https://flowrunner.ai/llms.txt