Positioning
What makes MyClaws different from a single-machine coding assistant?
The point is not a prettier chat box. It brings real hosts, real permissions and real workflows into the loop: multiple Claws, subagents, Trigger, mobile continuation and cross-conversation memory.
Do I need multiple machines for it to be useful?
No. Start with one Client and one Claw. As soon as you want internal-network access, a cloud runner, a home server or overnight execution, the value of multi-host orchestration shows up naturally.
Is it for solo developers or teams?
Both. Individual developers use it to glue several machines and multiple model providers together. Teams use it for on-call work, scheduled checks, cross-environment verification and multi-role collaboration.
Deployment and operation
Should I install Client first or Claw first?
Either works. The easiest path is to install Client first: Client has an Add Claw button that walks you through SSH credentials and installs Claw on the target host automatically β no command line required.
Can mobile replace desktop?
Not really. Mobile is best for continuity, acknowledgements, notifications and task boards. Dense debugging and deep review remain better on desktop.
Security, permissions and trust
Will MyClaws automatically isolate risky commands for me?
Not by wrapping everything in another sandbox. The product makes boundaries explicit instead: execution happens on a real host, while Agent / Skill / MCP / tool allowlists decide capability exposure.
Where do the trust boundaries actually sit?
- Admin handles authentication, configuration, downloads and push relay. It does not hide the host's real permission boundary.
- Claw runs on your real host using its files, network and toolchain β manage credentials and network egress like any serious automation system.
- Tool allowlists keep higher-risk capabilities (bash, write, web search, MCP) confined to the Agents that genuinely need them.
Can I start with low-risk monitoring only?
Yes β and that is usually the smoothest way in. Run Trigger, message center and mobile continuation on reversible tasks first, then expand. Suggested baseline: connect a low-permission host first; create role-specific Agents; separate desktop and server duties; keep critical review on desktop.
Commercial use
I want to use this commercially or partner β who do I contact?
Email the author directly: weidwonder@gmail.com. Whether you are exploring commercial deployment, custom integration or a partnership, drop a short note about your scenario and we can figure out the right path.