Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.rigbox.dev/llms.txt

Use this file to discover all available pages before exploring further.

The Rigbox app catalog provides curated, pre-packaged applications that you can install into any workspace with a single CLI command. Each catalog app is configured with the correct port, subdomain, and systemd service — no manual setup required.

How It Works

When you install a catalog app:
  1. Rigbox downloads and configures the app inside your workspace
  2. For Service-shape apps: a systemd service is created, an app route maps the port to a *.rigbox.dev subdomain, and a readiness probe waits for the service to come up
  3. For CLI-shape apps: the binary is dropped into $PATH (and any routing/profile.d shims are written), then the install returns — no systemd unit, no port, no route, no probe
  4. The app row is registered either way, so the workspace UI shows the app as installed
You manage installed Service-shape apps with the same endpoints you use for any other app — start, stop, restart, delete, and change visibility. CLI-shape apps don’t expose those endpoints; you invoke them interactively over SSH.

App shapes

ShapeWhat you getWhat’s missingTypical use
Servicesystemd unit, Caddy route, subdomain, health probeVS Code, Jupyter, Streamlit, PGWeb, …
CLIBinary on $PATH, optional profile.d routing shimsystemd, route, subdomain, port, probeclaude, codex, opencode, kilocode
For CLI-shape apps the install-app job is finished as soon as install.sh exits — there is no service readiness to wait on. Once the job status is completed, SSH into the workspace and run the agent directly. See AI Coding Tools for usage patterns once a recipe is installed, and Managed AI Proxy for how credits are metered for the requests each agent makes.
User-created apps (those you POST to /apps directly, instead of installing a catalog recipe) accept the same app_kind: "service" | "cli" field. CLI-shape apps must have port: 0 and omit protocol; Service-shape apps default to port 3000 and require a protocol.

Available Apps

List the catalog from any workspace or from your local machine:
rig catalog
See List Catalog Apps for the underlying HTTP shape.

Service-shape Catalog Reference

AppCatalog IDPortSubdomainDescription
VS Code Servervscode8443vscodeBrowser-based VS Code with full extension support
Jupyter Labjupyter8888jupyterInteractive notebooks for Python, Julia, R
Marimomarimo8000marimoReactive Python notebooks
Streamlitstreamlit8501streamlitPython data app framework for dashboards
File Browserfilebrowser8080filesWeb-based file manager with upload/download
Excalidrawexcalidraw3000excalidrawCollaborative whiteboard and diagramming
PGWebpgweb8081pgwebPostgreSQL web client with query builder
Subdomains are prefixed with the workspace name. For example, if your workspace is named my-workspace and you install the VS Code catalog app, the resulting app name is my-workspace-vscode and the full URL is https://my-workspace-vscode.rigbox.dev. Replace my-workspace with whatever name you gave your workspace.

CLI-shape Catalog Reference

These recipes install an interactive AI coding agent into the workspace’s $PATH. They route through the managed AI proxy (Rigbox’s OpenRouter gateway), so no provider API key is required when the workspace is in managed AI mode.
AgentCatalog IDInvoke asRouting notes
Claude CodeclaudeclaudeOPENROUTER_* injected; profile.d shim adds Anthropic-compatible env
Codex CLIcodexcodexOPENROUTER_* → translated to OPENAI_BASE_URL / OPENAI_API_KEY by profile.d
OpenCodeopencodeopencodeReads OPENROUTER_API_KEY natively — no env translation needed
Kilo CodekilocodekilocodeOPENROUTER_*KILO_OPEN_ROUTER_API_KEY by profile.d
Install one of them the same way as a Service-shape app:
rig catalog install --item claude --workspace <WORKSPACE_NAME>
Inside a workspace VM, drop the --workspace flag — rig infers it from the ambient identity. Then SSH in and run the agent — rig ssh-info <workspace> will print the right ssh command.
CLI-shape recipes don’t appear in the Start App / Stop App / visibility endpoints — there’s no daemon to control. To uninstall, delete the app row; the catalog’s cleanup paths (e.g. /etc/profile.d/codex-routing.sh) are removed automatically.

Installing a Catalog App

Install an app by its catalog ID:
rig catalog install --item vscode --workspace <WORKSPACE_NAME>
rig catalog install blocks until the install job completes and returns the app’s name and URL when it’s done. From inside a workspace VM, omit --workspace. For programmatic installs that need to fire-and-forget or track the job ID separately, use Install Catalog App and Get Job Status directly.

Access the installed app

Once installed, a Service-shape app is running and accessible at its subdomain URL. CLI-shape apps are invoked over SSH (see App shapes above).

Managing Installed Apps

After installation, catalog apps behave like any other app. Use the standard app endpoints to manage them.

Start and Stop

rig app start <APP_NAME>
rig app stop <APP_NAME>
rig app restart <APP_NAME>
See Start App, Stop App, and Restart App for the API form.

Change Visibility

Catalog apps are private by default. Visibility is managed through the API:
curl -X PUT https://api.rigbox.dev/api/v1/apps/$APP_ID/visibility \
  -H "Authorization: Bearer $RIGBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"visibility": "public"}'
See App Visibility for the full set of options and Update App Visibility for the API reference.

Delete

Deleting an installed app removes both the route and the systemd service inside the workspace:
rig app rm <APP_NAME>
See Delete App for the API form.

App Details

Each app installs with the same one-liner — swap in the catalog ID from the table above:
rig catalog install --item vscode --workspace <WORKSPACE_NAME>

VS Code Server

A full browser-based VS Code instance with extension support, integrated terminal, and file editing. Runs code-server.
VS Code Server supports installing extensions from the Open VSX registry. Search and install extensions directly from the browser UI.

Jupyter Lab

Interactive notebooks for Python, Julia, R, and more. Supports inline visualizations, markdown cells, and kernel management.

Marimo

Reactive Python notebooks where cells automatically re-execute when dependencies change. An alternative to Jupyter with a focus on reproducibility.

Streamlit

Build interactive data dashboards and web apps in Python. Write a Python script and Streamlit renders it as a web application.

File Browser

A web-based file manager with drag-and-drop upload, download, and file editing. Useful for managing files inside the workspace without SSH.

Excalidraw

A collaborative whiteboard for sketching diagrams, wireframes, and architecture drawings. Runs locally inside the workspace — no external dependencies.

PGWeb

A web-based PostgreSQL client with a query editor, table browser, and export functionality. Useful for inspecting databases running inside the workspace.

Image Compatibility

Most catalog apps work on all images, but some have dependencies that not every image provides.
Appbasedevfullopenclaw
VS Code ServerYesYesYesYes
Jupyter LabYesYesYesNo (needs Python)
MarimoYesYesYesNo (needs Python)
StreamlitYesYesYesNo (needs Python)
File BrowserYesYesYesYes
ExcalidrawYesYesYesYes
PGWebYesYesYesYes
The base image has pre-warmed npm caches for Jupyter, Marimo, Streamlit, and Excalidraw. Installation on base is significantly faster than on other images because dependencies are already downloaded.

Complete Example: Install VS Code and Jupyter

Install both into the same workspace:
rig catalog install --item vscode --workspace my-project
rig catalog install --item jupyter --workspace my-project
Each command blocks until the install completes and prints the app’s URL. After both finish, the apps are accessible at https://my-project-vscode.rigbox.dev and https://my-project-jupyter.rigbox.dev.

Next Steps