Runwork
All Features

Runwork CLI

Your terminal interface to the work cloud

One CLI to connect your AI agents, manage workspace resources, and develop apps. Zero local dependencies. Full platform parity with the web dashboard.

Install Runwork CLI

No runtime dependencies. Works on macOS, Linux, and Windows.

Open a terminal and paste.

Also available via

Direct download GitHub Releases

Then run runwork login to authenticate and runwork setup to connect your AI agents.

Get Started

Connect yourself and your AI agents to the workspace.

runwork login

Authenticate with the Runwork platform.

$ runwork login
✓ Authenticated as oytun@runwork.ai
✓ Workspace: MotaWord

runwork setup

Auto-detects and configures 8 AI agents with your workspace skills and MCP servers: Claude Code, Claude Desktop, Cursor, Codex, Windsurf, Copilot (VSCode + CLI), and Gemini.

$ runwork setup
Detected agents:
✓ Claude Code
✓ Claude Desktop
✓ Cursor
✓ Codex
✓ 12 skills synced
✓ 4 MCP servers configured
✓ Your agents are connected to MotaWord workspace

AI Ecosystem

Manage skills, MCP servers, agents, and keep everything in sync.

runwork skills

List workspace skills or push new ones from local files.

$ runwork skills list
App Skills:
CRM Dashboard 5 entities, 3 workflows
Invoice Manager 2 entities, 1 schedule
External Skills:
Sales Playbook imported from SKILL.md
$ runwork skills push ./my-skill.md
✓ Skill "content-guidelines" pushed to workspace

runwork mcp

Add, remove, and list MCP servers in your workspace.

$ runwork mcp list
runwork-workspace 12 tools enabled
hubspot-crm 8 tools enabled
google-analytics 5 tools enabled
$ runwork mcp add reporting https://api.example.com/mcp
✓ MCP server "reporting" added (6 tools discovered)

runwork agents

List agents, check status, and view execution history.

$ runwork agents list
data-analyzer CRM App active
report-writer CRM App active
support-bot Help Desk active
$ runwork agents executions
s-a8f3e2 data-analyzer completed $1.24 3m42s
s-c4d1b7 report-writer completed $0.67 1m15s

runwork sync

Refresh local agent configs with the latest workspace skills and MCP servers.

$ runwork sync
✓ 2 new skills synced
✓ 1 MCP server updated
✓ Claude Code config refreshed
✓ Cursor config refreshed

Workspace Resources

Create and manage work primitives from the terminal.

runwork entities

Full CRUD for workspace entity data. Create, read, update, delete records.

$ runwork entities create customers --data '{"name":"Acme Corp"}'
✓ Record created: cust_2x8f9a3
$ runwork entities records customers --limit 3
cust_2x8f9a3 Acme Corp 2 invoices
cust_1k4m7b2 TechStart Inc 5 invoices
cust_9p2n5d1 Global Media 1 invoice

runwork workflows

List, trigger, and inspect workflow instances.

$ runwork workflows trigger onboarding --params '{"customer":"Acme"}'
✓ Workflow started: wf_inst_3k2m
$ runwork workflows instances --app "CRM"
wf_inst_3k2m onboarding running 12s
wf_inst_1a9b follow-up completed 45s

runwork schedules

Manage scheduled jobs: trigger, pause, resume, view history.

$ runwork schedules list
daily-digest 0 9 * * * active CRM App
weekly-backup 0 2 * * 1 active Billing
hourly-sync 0 * * * * paused Data Sync
$ runwork schedules resume hourly-sync --app "Data Sync"
✓ Schedule "hourly-sync" resumed

runwork integrations

Search the catalog, list connections, and call integration APIs directly.

$ runwork integrations call hubspot GET /crm/v3/contacts?limit=2
200 OK
{ "results": [
{ "firstname": "Sarah", "lastname": "Chen" },
{ "firstname": "James", "lastname": "Wu" }
] }

runwork files

Upload, download, and manage files in workspace storage buckets.

$ runwork files upload reports ./q2-summary.pdf
✓ Uploaded q2-summary.pdf (2.4 MB)
$ runwork files ls reports
q2-summary.pdf 2.4 MB Mar 25
q1-summary.pdf 1.8 MB Jan 15

runwork endpoints

List public endpoints and call them directly from the terminal.

$ runwork endpoints list
GET /api/customers CRM App api-key
POST /api/webhooks Billing none
$ runwork endpoints call GET /api/customers
200 OK [3 records]

Development

Build and ship vibe apps. Ideally your AI agent handles this for you.

runwork init

Create a new app. The platform provides CLAUDE.md + AGENTS.md on the fly, plus SKILL.md for your app.

$ runwork init "Lead Tracker"
✓ App created in ./lead-tracker
✓ Project skeleton ready
→ Run `runwork dev` to start developing

runwork dev

Boots a remote sandbox with live preview and file watching. No Docker, no local runtime.

$ runwork dev
✓ Sandbox ready
✓ CLAUDE.md + AGENTS.md synced
✓ SKILL.md generated
✓ Preview → https://preview-a3x.runwork.ai
[HMR] entities.ts changed — rebuilding (0.4s)

runwork deploy

One command to production. No build pipelines to configure.

$ runwork deploy
✓ Code pushed
✓ Building...
✓ Deployed to production
→ https://your-workspace.runwork.ai

Access from anywhere

Claude ChatGPT Cursor

AI Agents

Ask your AI agent:

Connect your AI agent

Web Dashboard

Every CLI command has a dashboard equivalent. Use whichever you prefer, or both.

Open in dashboard

CLI

Use Cases

CLI-first developers who live in the terminal Claude Code users building full-stack apps Cursor users who want seamless cloud deployment Codex users automating app creation Teams who prefer terminal workflows over GUIs

Why It Matters

How It Works

Connect your AI agents. Run runwork setup to auto-detect and configure up to 8 AI agents, including Claude Code, Claude Desktop, Cursor, Codex, Windsurf, Copilot (VSCode and CLI), and Gemini, with your workspace's skills and MCP servers. Run runwork sync whenever your workspace changes. Your agents always have the latest context.

Manage workspace resources. Create entities, trigger workflows, pause schedules, upload files, call integrations, manage API keys. Every workspace feature is a CLI command. Lead with creation: runwork entities create, runwork workflows trigger, runwork schedules trigger.

Develop and deploy apps. runwork init creates a new app. runwork dev boots a remote sandbox with live preview and file watching. runwork deploy ships to production. During development, the platform provides CLAUDE.md + AGENTS.md on the fly, plus auto-generated SKILL.md per app. Claude Code, Cursor, and Codex understand your project from the moment you start.

Zero local dependencies. When you run runwork dev, your code executes in a remote sandbox. You edit files locally, the CLI syncs them via git protocol, and the sandbox handles builds, execution, and preview serving. No Docker, no Node, no runtime on your machine.

Ideally, your AI agent handles all of this for you. You say "build me a customer portal" and your agent runs init, dev, deploy under the hood. The CLI is also the bridge that powers agent sandboxes, giving your AI agents the same terminal capabilities you use during development.

Frequently Asked Questions

Do I need the CLI to use Runwork?
No. The web dashboard provides full access to every Runwork feature. The CLI is an additional option for developers who prefer working in the terminal or want to integrate Runwork into their existing editor and AI tool workflows.
What are the system requirements?
The CLI runs on macOS (Apple Silicon and Intel), Linux (x64 and ARM), and Windows. It is distributed as a standalone binary with no runtime dependencies. You do not need Node.js, Docker, or any other toolchain installed.
How does local development work without local dependencies?
When you run runwork dev, the CLI syncs your local files to a remote sandbox via git protocol. The sandbox handles compilation, execution, and serves a live preview. You edit files in your preferred editor, save them, and see results instantly. No local runtime, database, or container engine required.
Can I use the CLI with Claude Code, Cursor, or Codex?
Yes. When you run runwork dev, the platform creates CLAUDE.md + AGENTS.md on the fly with full framework context, and SKILL.md is auto-generated for your app. These knowledge files are read automatically by Claude Code, Cursor, Codex, and other AI coding tools. You can also download SKILL.md from the web dashboard for any app.
Is the CLI open source?
The CLI is distributed as a compiled binary. The @runworkai/framework npm package is available separately and provides TypeScript type definitions for building Runwork apps with full editor support and intellisense.

Related Features

See How Teams Use Runwork CLI

Ready to try Runwork CLI?

Try the work cloud for AI agents.