Legal
Terms of Service
Draft — last updated August 7, 2026.
1. Agreement to these terms
2. What Workser provides
Workser Business lets you build web apps, workflow automations, and AI Employees (autonomous agents that operate software and data on your behalf) inside a project-based workspace with its own database, storage, and hosting.
Workser Desktop is a desktop application that gives an AI agent computer-use capabilities you explicitly approve (screen viewing, mouse/keyboard control, file and browser automation), and includes Orbit, a bring-your-own-key bridge between the coding-agent CLIs you already run (currently Claude Code, Codex, OpenCode, Cursor, and Muse Code) and Workser’s deploy/hosting infrastructure.
3. Accounts
4. Acceptable use
Full detail is in our standalone Acceptable Use Policy. In summary, you agree not to use the Services to:
- Violate applicable law, or infringe anyone’s intellectual property, privacy, or other rights.
- Attempt to bypass or interfere with security controls, sandbox isolation, or access-control mechanisms.
- Use an AI agent to attempt prompt injection, data exfiltration, or unauthorized access to systems you don’t own or haven’t been authorized to access.
- Use computer-use or Orbit capabilities to take actions on a device or account you’re not authorized to control.
- Resell or provide the Services to third parties without our prior written consent, except as expressly permitted by your plan.
5. Your content and AI output
6. AI agents and autonomous actions — please read carefully
Workser’s AI Employees and computer-use agents can take real actions on your behalf — writing and running code, calling connected third-party apps, controlling your computer, and more. We build in safeguards (sandboxed execution, secret redaction, prompt-injection guardrails, and approval prompts for computer-use actions), but AI systems can make mistakes.
You’re responsible for reviewing and approving agent actions where the Services request your approval, for configuring appropriate limits (e.g. per-user credit limits, connected-app scopes) for your organization, and for the consequences of actions your agents take within the scope you’ve granted them. Don’t grant an agent access to systems or data where an error would cause harm you’re not prepared to accept.