Install
Install the skipprd command, the skippr Python package, or both. They run the same engine, so a skippr.yml written by one works with the other.
| You want to | Install |
|---|---|
| Run pipelines from a terminal, cron, or a container | skipprd with Homebrew or install.sh |
| Build and run pipelines from Python or a notebook | skippr with pip |
Installing Skipprd means accepting the Skipprd EULA.
Before you begin
- A macOS arm64 (Apple silicon) or Linux x86_64 machine. Other platforms are not published yet.
- For
pip: Python 3.10 or later. - HTTPS access from that machine to
install.skippr.io. Skipprd downloads each connector the first time a pipeline uses it. - Network access from that machine to your source and destination.
Install the CLI
brew tap skipprd/tap
brew install skipprdcurl -sL https://raw.githubusercontent.com/skipprd/skipprd/main/install.sh | shHomebrew is the simplest path on a Mac. It also works on Linux when brew is on your PATH.
The install script detects your platform, downloads the latest release from install.skippr.io, and writes skipprd to /usr/local/bin. If that directory is not writable it asks for sudo. To install without root, choose a directory on your PATH:
curl -sL https://raw.githubusercontent.com/skipprd/skipprd/main/install.sh | SKIPPR_INSTALL_DIR="$HOME/.local/bin" shTo pin a release, for example on a production host, set SKIPPR_VERSION to an unprefixed version number:
curl -sL https://raw.githubusercontent.com/skipprd/skipprd/main/install.sh | SKIPPR_VERSION=6.10.0 shYou can also download the Linux x86_64 or macOS arm64 archive from GitHub Releases and put skipprd on your PATH yourself.
Install the Python package
pip install skipprThe package includes type stubs, so your editor and mypy see every connector class and field. It depends on pyarrow; query results come back as Arrow tables. For pandas conversion, install the extra:
pip install "skippr[pandas]"The Python package does not add the skipprd command. Install the CLI as well if you want both.
Verify the install
skipprd --versionpython -c "import skippr; print(skippr.Session)"skipprd --version prints the version number. The Python check prints <class 'skippr.Session'>.
Connectors
Connectors are not bundled with the engine. The first time a pipeline uses a source or destination, Skipprd downloads that connector from install.skippr.io into the connector cache (~/.skippr/runtime_plugins). Later runs use the cache.
To keep the cache somewhere else — for example, on a persistent volume in a container — set SKIPPR_RUNTIME_PLUGIN_DIR:
export SKIPPR_RUNTIME_PLUGIN_DIR="/var/lib/skippr/connectors"Upgrade
brew upgrade skipprdcurl -sL https://raw.githubusercontent.com/skipprd/skipprd/main/install.sh | shpip install --upgrade skipprTroubleshooting
skipprd: command not found— the install directory is not on yourPATH, or you installed only the Python package. Runecho $PATHand add the directory you installed to (for example,export PATH="$HOME/.local/bin:$PATH"), or install the CLI with Homebrew orinstall.sh.macOS x86_64 release assets are not published yetorLinux arm64 release assets are not published yet— the installer found an unsupported CPU. Run Skipprd on an Apple silicon Mac or an x86_64 Linux machine.Windows detected— Windows builds are not published. Run Skipprd on a macOS or Linux machine.release version must be unprefixed semver—SKIPPR_VERSIONincludes a prefix. Use6.10.0, notv6.10.0.Download did not return a tar.gz archive— no release exists for that version and platform. Check the version number, or unsetSKIPPR_VERSIONto install the latest release.unsupported SKIPPR_BINARY— this installer only installsskipprd. UnsetSKIPPR_BINARY.pipcannot find a matching distribution — wheels are published for macOS arm64 and Linux x86_64 on Python 3.10 or later. Checkpython --versionand your platform.- A pipeline hangs or fails on first run while fetching a connector — the machine cannot reach
install.skippr.ioover HTTPS. Test from the machine that runs the pipeline:curl -I https://install.skippr.io.
Next steps
- Quickstart: S3 to Athena — land sample data in a queryable AWS table.
- Quickstart: Snowflake, PostgreSQL, or BigQuery.
- Python — build and run pipelines from code.
- How Skipprd works — what happens to your data between source and destination.
