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.

Every workspace runs on a base image - a pre-built Debian 12 root filesystem with tools and runtimes already installed. Templates combine an image with default resource settings (RAM, vCPU, disk) so you can deploy in one call.

​
Images

Images determine what software is available inside your workspace out of the box. All images include SSH access, systemd, and the Rigbox CLI.
ImageSizeIncludesBest for
base3 GBPython 3.12, Node.js 22, uv, git, vim, tmux, GitHub CLIGeneral development, scripts, web apps
dev4 GBEverything in base + Docker CLI, cmake, build tools, strace, gdb, PostgreSQL clientSystems programming, native builds
full5 GBEverything in base + Architecture Explorer, Virtual Browser pre-installedFull tooling, code analysis
openclaw2.5 GBNode.js 22, OpenClaw gateway pre-installedAI bot deployment (Telegram, web)
All images are built on Debian 12 (Bookworm) and include Python 3.12 and Node.js 22 except where noted. The openclaw image is specialized and does not include Python.

​
What’s in each image

base is the right choice for most workloads. It includes:
  • Python 3.12 with uv for fast package management
  • Node.js 22 with npm
  • git, vim, tmux, curl, wget, jq
  • GitHub CLI (gh)
  • Pre-warmed npm caches for popular tools (Jupyter, Marimo, Streamlit, Excalidraw)
dev adds build tooling on top of base:
  • Docker CLI (for building images - Docker daemon runs on the host)
  • cmake, make, gcc, g++
  • strace, gdb, ltrace for debugging
  • PostgreSQL client (psql)
full adds pre-installed Rigbox tools:
  • Architecture Explorer - ready to launch without installation
  • Virtual Browser (Chromium + Firefox via Playwright) - ready to launch
  • Xvfb and x11vnc for headless display
openclaw is a minimal image for AI bot workloads:
  • Node.js 22 with the OpenClaw gateway pre-configured
  • Designed for Telegram bots, web chat bots, and similar agents

​
Listing Available Images

Images are not exposed as a standalone API. The set of images you can boot is implied by the templates the server publishes — every template’s image field names the base image it uses. To see which images are currently available, list templates and read the image field: 
curl -s https://api.rigbox.dev/api/v1/templates \
  -H "Authorization: Bearer $RIGBOX_API_KEY" | jq '[.templates[].image] | unique'
See List Templates for the full response schema.

​
Templates

Templates are pre-configured workspace definitions that pair an image with sensible resource defaults. Use them with quick deploy for one-call workspace creation. Requests below a template’s RAM, vCPU, or disk floor are raised before the workspace is created or started.
TemplateImageRAMvCPUDiskUse case
basebase512 MB13 GBMinimal workspace for scripts and web apps
devdev1 GB14 GBDeveloper environment with build tools
fullfull1 GB15 GBFull tooling pre-installed (Architecture Explorer, Virtual Browser)
openclawopenclaw1 GB12.5 GBAI bot workspace with OpenClaw gateway
ai-agent-starterbase1 GB23 GBAI sandbox with Open Terminal for agent workloads
custom-repobase1 GB13 GBClone and run a git repository

​
Listing Available Templates

rig template ls
See List Templates for the full response schema.

​
Choosing the Right Image

Use this decision tree to pick the image that fits your workload:

​
Do you need AI bot deployment?

If you’re building a Telegram bot, web chat agent, or similar AI-powered service, use the openclaw image. It has the OpenClaw gateway pre-installed.

​
Do you need Architecture Explorer or Virtual Browser?

If you want to visualize code structure or run a remote browser without installing anything, use the full image. Both tools are pre-installed and ready to launch.

​
Do you need Docker, cmake, or native build tools?

If you’re compiling C/C++ code, building Docker images, or need debugging tools like strace and gdb, use the dev image.

​
For everything else, use base

The base image covers Python, Node.js, and general development. It’s the smallest and fastest to boot.
You can always install additional packages inside any workspace with apt-get. The image choice just determines what’s pre-installed for faster startup.

​
Quick Deploy with a Template

The fastest way to create a workspace is rig workspace spawn with a template ID: 
rig workspace spawn --name my-project --template dev
rig workspace spawn creates the workspace, boots it, and waits until it’s ready - all in one call. See Quick Deploy for the API form.

​
Custom Workspace with a Specific Template

If the template defaults don’t fit your needs, create a workspace manually and specify the template along with custom resource settings. 
rig workspace new --name heavy-compute --template dev --ram 2048 --vcpu 2 --disk 8192
See Create Workspace for the API form.

​
Resizing After Creation

You can change the RAM, vCPU, and disk size of an existing workspace without switching images. The workspace must be stopped first. 
rig workspace stop --workspace my-project

curl -X PUT https://api.rigbox.dev/api/v1/workspaces/$WORKSPACE_ID/resources \
  -H "Authorization: Bearer $RIGBOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"ram_mb": 2048, "vcpu_count": 2}'

rig workspace start --workspace my-project
You cannot change the template of an existing workspace. To switch templates, create a new workspace with --template set to the one you want.
See Update Workspace Resources for details.

​
Pre-warmed Caches

The base image includes pre-warmed npm caches for several popular tools. This means installing catalog apps like Jupyter, Marimo, Streamlit, and Excalidraw is significantly faster because dependencies are already downloaded.
ToolCache benefit
Jupyter Labnpm packages cached, installs in ~30 seconds
Marimonpm packages cached, installs in ~20 seconds
Streamlitpip packages cached via uv
Excalidrawnpm packages cached, installs in ~15 seconds
See Catalog Apps for how to install these tools.

​
Next Steps