Skip to content

Guide · Open source

How to use Quivr: install, run locally and search content

Updated · written with AI from the project's own documentation and checked against our catalogue data

Quivr is an open-source engine for content that keeps arriving. You send it documents, or let it collect them from RSS feeds, Microsoft 365 mailboxes and X lists, and it makes them searchable and checks them against alerts you save. This guide shows you how to start it locally, try search and alerts, and decide whether it fits a developer project.

QuivrThe-Vibe-Company/quivr · ★ 39.6k · MIT

Quick start

  1. 1Check your system. Confirm you run Linux on x86_64 or a Mac with Apple Silicon, and that Docker with Compose v2, Python 3.12 with venv, and jq are installed.
  2. 2Clone the repository. Run the git clone command from the README and change into the quivr folder.
  3. 3Start Quivr. Run make dev to start Quivr and its dependencies in Docker.
  4. 4Point your shell at the API. Run eval with make -s env so your shell can reach the API with a local key.
  5. 5Try the quickstart. Create a collection, add an article, and search for it by keyword and by meaning.
  6. 6Set an alert. Save an alert on a topic and check that matching documents are recorded with the reason.

What Quivr does and who it is for

Quivr stores every version of each document, makes it searchable within seconds, and records new documents that match a topic you follow. Your application talks to it over an HTTP API. Quivr has no user interface of its own, so the quickstart and the optional demo web app are the only ready-made ways to see it working.

It suits developers building a search or monitoring pipeline over a stream of documents, and teams evaluating a self-hosted, plugin-based ingestion and search engine. It is not a ready-made chat or end-user app.

  • Collect documents one by one, in batches, or pulled from sources on a schedule
  • Search by keywords, by meaning, or both, with results pointing to the exact passage and version
  • Save alerts as a keyword expression or a plain-language description
  • Extend file formats, sources, embedding models, ranking and alert rules with plugins

Check the requirements before you install

Quivr runs on Linux on x86_64 or on a Mac with Apple Silicon. Windows and Intel Macs are not listed as supported. You also need Docker with Compose v2, the Go version named in the repository's go.mod file, Python 3.12 with venv, and jq. The first start downloads about 1 GB.

  • Linux x86_64 or macOS on Apple Silicon
  • Docker with Compose v2
  • The Go version listed in go.mod
  • Python 3.12 with venv
  • jq

Install and start Quivr locally

Quivr is installed by cloning the repository and building it, not from a package manager. Clone the repository, then start Quivr and its dependencies in Docker. The second command points your current shell at the API using a local key.

make dev starts Quivr and its dependencies in Docker. The quickstart then walks you through creating a collection, adding an article, and searching for it.

git clone https://github.com/The-Vibe-Company/quivr.git && cd quivr make dev # start Quivr and its dependencies in Docker eval "$(make -s env)" # point this shell at the API, with a local key

Try search, alerts and the demo web app

Follow the Quickstart on the documentation site. It covers creating a collection, adding an article, finding it by keyword and by meaning, and getting alerted on a topic, in about fifteen minutes.

If you want a visual front end on top of the API, make demo starts a web app at http://127.0.0.1:5183. Its code lives in the quivr-search folder.

  • Create a collection, add an article, then search by keyword and by meaning
  • Save an alert and check that new matching documents are recorded with the reason
  • Use the demo web app to browse results if you prefer a screen over API calls
make demo

Run the checks and contribute

Run make check before pushing changes. It takes about two minutes and needs no Docker. make verify runs the end-to-end journeys on an isolated stack.

Read AGENTS.md for the rules and CONTEXT.md for the domain vocabulary before contributing. Pull request titles follow Commitizen, for example feat(ingestion): accept record versions. Report security issues privately, as described in SECURITY.md.

make check

Know the current status before you rely on it

Quivr 2 keeps the HTTP API under /v0 compatible across 2.x releases, and upgrades keep your data in place, according to the README. The project analysis describes Quivr as being at an evaluation stage and not yet ready for production, and the API is described as v0 that may change. Read the Known limits page before building anything you cannot afford to rework.

What people use it for

  • Monitoring news articles or posts for a topic and reading matches through the API
  • Building a searchable archive of emails or PDFs that keeps every version of each document
  • Running a self-hosted search pipeline with custom file-format or ranking plugins
  • Letting an AI agent search Quivr and quote passages with their sources using quivr mcp

Frequently asked questions

Is Quivr free to use?

Quivr is released under the MIT licence, as stated in the README and the repository's LICENSE file. The source does not list any paid tier, so you install and run it yourself.

What are the requirements to run Quivr?

You need Linux on x86_64 or a Mac with Apple Silicon, Docker with Compose v2, the Go version in go.mod, Python 3.12 with venv, and jq. The first start downloads about 1 GB.

Does Quivr work on Windows?

The quickstart lists only Linux on x86_64 and macOS on Apple Silicon. Windows and Intel Macs are not listed as supported in the source.

Is Quivr ready for production?

The project analysis says Quivr is at an evaluation stage and not yet ready for production. The README says the API is v0 and may change, so check the Known limits page before depending on it.

Does Quivr have a user interface?

No. The README says Quivr has no user interface of its own and is used through an HTTP API. A demo web app is available with make demo for trying it out.

How does Quivr compare to Elasticsearch or Meilisearch?

Elasticsearch, Meilisearch, Typesense and Quickwit are listed as comparable projects. Quivr differs in focus, since it is built around continuous ingestion, versioned documents, alerts and plugins, but this guide does not benchmark them.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Quivr suits developers who want to evaluate a self-hosted, plugin-based engine for storing, searching and alerting on a stream of documents, and who are comfortable with Docker Compose and API docs. Pick something else if you need a production-ready system now, a ready-made chat or end-user app, or you work on Windows or an Intel Mac.

Install options & downloads