CLI
Command reference
Sandwalk CLI subcommands and interactive slash commands.
CLI Subcommands
sandwalk setup # providers, keys, default model
sandwalk desktop # install the Mac app or Ubuntu .deb
sandwalk # interactive session
sandwalk "question" # one-shot query
sandwalk serve # local agent API on :8787
sandwalk serve --host 0.0.0.0 # expose to the LAN (needs sandwalk auth)
sandwalk auth # manage the API key for serve
sandwalk login chatgpt # use a ChatGPT Plus/Pro plan for OpenAI models
sandwalk doctor # environment and configuration checks
sandwalk eval # run Benchmarking suites headlessly
sandwalk cron list # manage scheduled Automations & cron workflows
sandwalk ssh list # manage SSH hosts, auto-install, and tunnels
sandwalk skills add fastfold-ai/skills
sandwalk mcp list
sandwalk sandbox status # Local Sandbox or Modal Sandbox
sandwalk data pull-all # DepMap, MSigDB, STRING
sandwalk tool list
sandwalk upgrade| Command | Description | Guide |
|---|---|---|
sandwalk | Interactive terminal REPL | Terminal |
sandwalk "…" | One-shot query for scripts & CI | One-shot |
sandwalk serve | Local agent API (:8787) | Serve |
sandwalk setup | First-time provider & model wizard | Quick Start |
sandwalk desktop | Download & install macOS .dmg or Ubuntu .deb | Desktop App |
sandwalk login chatgpt | Sign in with ChatGPT Plus/Pro OAuth | Models |
sandwalk eval | Run Benchmarking suites headlessly | Benchmarking |
sandwalk cron … | Create, list, pause, resume, or trigger scheduled Automations | Automations |
sandwalk ssh … | Manage SSH hosts, remote commands, and 1-click Remote-SSH bootstrap | SSH & Remote Servers |
sandwalk doctor | Environment & readiness checks | Doctor |
sandwalk config show|set | Inspect or update ~/.sandwalk/config.json | Config |
sandwalk tool list | Inspect the 190+ tool registry | Tools |
sandwalk skills … | Find, add, upgrade, or list skills | Skills |
sandwalk mcp list | List configured MCP servers | MCPs |
sandwalk data … | Pull local corpora (DepMap, MSigDB, STRING) | Data |
sandwalk sandbox … | Local vs. Modal CPU/GPU sandbox helpers | Sandbox |
Interactive Slash Commands
Type / inside sandwalk to open the command palette:
| Command | What it does |
|---|---|
/models, /effort | Switch the model; set Low / Medium / High / Fast for models with thinking controls |
/plan | Toggle Plan mode (plans/*.md + Mermaid flowchart) |
/sandbox | Pick Local or a Modal CPU/GPU preset; edit the base image inline |
/specialist | Use, clear, or list Specialists (including starter personas & custom profiles) |
/skills, /skills-add, /skills-find | Loaded skills; install from GitHub or a path; search the catalog |
/mcps | Connect, enable, validate, add, or remove MCP servers |
/tools, /data | Tool status; local datasets |
/usage, /compact | Token and cost for the session; compress context |
/sessions, /resume, /new | List, resume, or start sessions |
/export, /notebook | Export transcript to Markdown; export session to .ipynb |
/agents | Run a question across N parallel agents |
/doctor | Readiness diagnostics with fix hints |
Last updated on