OngoingAIOngoingAI Docs

Install

Use this page to install OngoingAI Gateway and verify the binary works.

Pick an installation path

Start with the install script unless you need containerized runtime boundaries or custom build outputs.

  • Use the install script for the fastest setup on a local machine.
  • Use Docker when you want a container runtime.
  • Build from source when you need to contribute or customize builds.

Prerequisites

  • For script install, you must have curl.
  • For Docker install, you must have Docker.
  • For source builds, you must have Go 1.24+ and make.

Install with the script

  1. Run the installer.

    Bash
    curl -fsSL https://ongoing.ai/install.sh | sh

  2. Verify the installed binary.

    Bash
    ongoingai version

    You should see a version string.

If you need to pin a version or change install behavior, set these variables before step 1:

VariablePurpose
ONGOINGAI_VERSIONInstall a specific release instead of latest.
ONGOINGAI_INSTALL_DIRInstall to a custom directory instead of /usr/local/bin.
ONGOINGAI_INSTALL_BASE_URLDownload from a custom release base URL.

Install with Docker

  1. Pull the image.

    Bash
    docker pull ghcr.io/ongoingai/gateway:latest

  2. Verify the container image.

    Bash
    docker run --rm ghcr.io/ongoingai/gateway:latest version

    You should see a version string.

Build from source

  1. Clone the repository and build.

    Bash
    git clone https://github.com/ongoingai/gateway.git
cd gateway
make build

  2. Verify the built binary.

    Bash
    ./bin/ongoingai version

    You should see a version string.

The build writes the binary to ./bin/ongoingai.

Build release targets

If you need release binaries for multiple platforms, run:

Bash
make build-cross

Build outputs are written to ./dist/.

Default runtime behavior

After install, the gateway uses these defaults:

  • Storage driver: SQLite at ./data/ongoingai.db
  • Provider route prefixes: /openai and /anthropic
  • Body capture: off (tracing.capture_bodies=false)

See the configuration reference for all available settings.

Troubleshooting

ongoingai command is not found after script install

  • Symptom: shell returns command not found: ongoingai.
  • Cause: install directory is not in your PATH.
  • Fix: add the install directory to PATH (default: /usr/local/bin), then start a new shell session.

Script install fails with a permission error

  • Symptom: installer cannot write the target binary.
  • Cause: your user cannot write to the install directory.
  • Fix: set ONGOINGAI_INSTALL_DIR to a writable directory, or install with a user that can write to /usr/local/bin.

Source build fails with a Go toolchain error

  • Symptom: make build fails with version or module errors.
  • Cause: local Go version is below 1.24.
  • Fix: upgrade Go, then rerun make build.

Next steps

PreviousWhat Is OngoingAI GatewayNextQuickstart