Quickstart

Open your workspaceCopy Anchor Link

Download and install OpenFactory on your computer. The workspace runs locally; this website hosts the downloads and documentation.

For a local checkout, use Node 26 and pnpm 12.15.0:

pnpm install
pnpm runtime:build
pnpm runtime:start
# In another terminal:
pnpm dev

Open your local workspace. Install and sign in to Codex CLI or Claude Code CLI on the computer running the runtime. OpenFactory reuses that authentication; T3 Code is not required. Keep the workspace on a trusted network, such as a private tailnet.

Install the terminal CLICopy Anchor Link

Open the Terminal section of the download page and copy its install command. The command downloads /install.sh from this website and installs the standalone openfactory CLI on macOS or Linux. It includes Node, so you do not need to install Node or npm yourself.

After installation, run:

openfactory --version
openfactory --help
openfactory serve

Keep openfactory serve running. In another terminal, use openfactory status to check the runtime. The CLI installer installs the runtime only. Use the desktop downloads for the graphical workspace. Install and authenticate Codex CLI or Claude Code CLI separately to execute agents.

By default, the installer stores versions in ~/.local/share/openfactory/runtime/versions and links the command into ~/.local/bin. If that directory is not on your PATH, the installer tells you to add it. Downloads are verified against the release's SHA256SUMS. Rerun the install command to get the latest published version. A failed download or checksum mismatch leaves the existing command unchanged.

Choose a factoryCopy Anchor Link

Select a project under Factories in the sidebar. OpenFactory lists projects from the OpenFactory runtime, including projects without scheduled agents.

The factory dashboard shows the activity available for that project. Use Agents to see its scheduled agents and Runs to inspect execution history.

Inspect an agentCopy Anchor Link

Open an agent card. Review its instructions, model, schedule, and recent runs before starting work.

Use Runs & assistant to read a previous run or continue the live conversation. Historical transcripts are read-only; switch to the live assistant to send a new message.

Start a runCopy Anchor Link

Choose Run now on an agent you want to execute. This dispatches real work through the connected runtime.

Open the resulting run or conversation to read the outcome. Dispatch success confirms that the runtime accepted the request, rather than that the agent finished its task.

Manage scheduled workCopy Anchor Link

Pause an agent to stop future scheduled dispatches, or resume it to enable them again. Timer schedules use the runtime host's timezone and require that host to remain running.

Continue with How factories work for the workspace concepts, or the factory definition reference for configuration details.