Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Personal cloud framework.
Shell Erlang Perl CoffeeScript JavaScript
Fetching latest commit...
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

Build Status

zeph·yr noun

  • a breeze from the west
  • a lightweight fabric

Zephyr is a lightweight personal cloud framework that makes it easy to build scalable, real-time apps powered by personal data.

Personal cloud is a relatively new term. In concept, it's a server that provides authentication and access to a user's personal data. Think of Facebook and Dropbox as proto-personal clouds. Each one provides some form of authentication and an API for fetching and storing data on behalf of the user. The problem with existing proto-personal clouds is that they are neither standardized nor portable which means it's not ideal for users and more work for app developers to integrate.

Learn more about Personal Clouds.

The goal of Zephyr is to create a simple personal cloud framework that provides standardized APIs for authenticating, fetching, normalizing, storing, and sharing data between personal clouds and within a scalable app architecture.

Support is being added to Zephyr for both existing personal data sources (Facebook, Dropbox, Gmail, etc.) as well as fully compliant personal clouds.

Zephyr is designed to be used either as your app's backend or as a personal data cache layer for your existing apps.


Zephyr is a new project but moving quickly to solve real world app developer problems.

  • Router – config and basic router [done]
  • CloudStore – token based authentication, GET, PUT, DELETE [done]
  • Tests – basic coverage for CloudStore [done]
  • Example Apps – Chat [done], data browser [WIP]
  • 3rd Party API Support
    • Singly for fetching social and personal data [WIP]
    • remoteStorage for unhosted apps [WIP]
    • connectors for Dropbox, Facebook, Twitter, Gmail, etc. [not started]


Zephyr is composed of several modular components that interact via HTTP APIs.


  1. Router: API endpoint that routes requests to other components
  2. CloudStore: key/value data store with OAuth security layer and notification events
  3. Pub/Sub: webhook callback engine for distributed event notification
  4. Engines: pluggable components to fetch social data, provide additional services, etc.

Each component is defined by a RESTful API and is completely replaceable by anything that provides the same API. Most components run behind a firewall and talk to each other on the internal network. Components can be spread out across any number of servers or run on a single server.


Zephyr runs on OSX and Linux. Windows has not yet been tested.


OSX Quick Install

brew install erlang
brew install postgres
brew install nodejs

# see Install Guide for postgres pg_hba.conf issues

Install Guide for Linux and detailed instructions


# make sure postgres is already running

git clone
cd zephyr
make setup
script/zephyr start
make test


Configuration options are in config directory files. When an option is changed, recompile and restart services.

# reapply config and recompile
# restart services
script/zephyr restart

Config Guide for options and details


# start console in foreground
script/zephyr console

# start as service
script/zephyr start

# stop service
script/zephyr stop

# restart service
script/zephyr restart


When Zephyr starts, it runs two data services:

  1. Data API
  2. Token Service

The Data API provides RESTful access to data. The Token Service allows for the creation and deletion of access tokens. For security reasons, the Token Service listens on a different port and (optionally) network interface.

Add Auth Token

An auth token is required to access data.

curl --request PUT --header "Content-Type: application/json" --data "{\"\":\"rw\"}" --verbose

In browser or curl, GET*?token=SECRET which should return an empty JSON object.

The token can be passed in as a cookie or query param.


curl --data "{\"foo\":\"bar\"}" --request PUT --header "Content-Type: application/json" --verbose


curl --verbose


curl --request DELETE --verbose

Note that PUT is a merge operation for objects. This makes it easy for clients to update specific attributes of an existing object without needing to send the entire object.

curl --data "{\"foo\":\"bar\"}" --request PUT --header "Content-Type: application/json" --verbose
curl --data "{\"new\":\"bar\"}" --request PUT --header "Content-Type: application/json" --verbose
curl --verbose

# Returns:
# {
#   "foo": "bar",
#   "new": "bar"
# }

To replace the entire object, use REPLACE or do a DELETE then PUT.

Data API Documenation


API tests are written in Node.js mocha

# run tests
make test

Zephyr must be already running for the tests to pass. Tests are run against the same database as development but on a guaranteed unique test branch that is deleted at the end of the tests.

Testing Guide.


API and Reference documentation are in the wiki.


Erlang compilation errors (usually rebar errors) are typically caused by having the wrong version of Erlang/OTP installed. Refer to Troubleshooting for help.


Join the Google+ Community or submit a pull request.

We especially appreciate hearing about your app ideas and how we can help make your life easier as a developer.

Something went wrong with that request. Please try again.