Home battery and energy apps
Read Powerwall, battery, or inverter status from mobile-only apps and trigger actions when thresholds are crossed.
Many services don't have APIs…
But they do have apps!
Clawperator
YOUR AGENT THINKS.
CLAWPERATOR ACTS.
Clawperator makes Android apps programmable - think Playwright for mobile apps.
Clawperator is a deterministic execution layer that lets agents run skills on real Android apps to read data and take action on behalf of users.
🧠 Your OpenClaw/agent is the brain.
✋ Clawperator is the hand.
Connect a dedicated Android burner phone, run your own private skills, and get answers your agent can trust.
Any cheap Android phone works. Log in to the accounts you care about and leave your burner ready for your agent. You do not need to switch from iPhone to Android.
Don't have an Android device? No problem. Clawperator can create a Google Play-equipped Android emulator for you.
What you can do with Clawperator
Clawperator is most useful where the real interface only exists in a mobile app.
Google Home and thermostat control
Check temperature, switch device state, and turn heating or cooling on without waiting for a public API that may never exist.
School app notification filtering
Monitor a noisy school app, filter the clutter, and surface only the updates that actually matter.
Family location flows
Check Life360-style apps and send updates or screenshots until someone gets home, then stop automatically.
Shopping, delivery, and status checks
Build private workflows around grocery, courier, or account apps that only really exist as mobile interfaces.
App-only balances and alerts
Pull statuses, balances, or alerts from user-installed Android apps that do not expose a usable public API.
Quick Start
One command installs the CLI, fetches the latest Clawperator Operator Android app, verifies it, and helps prepare an Android device for your agent.
# Install Clawperator on macOS/Linux
curl -fsSL https://clawperator.com/install.sh | bashNo Android device handy? Have Clawperator create a Google Play equipped Android emulator.
clawperator provision emulatorClawperator has comprehensive documentation, setup guides, and API references at docs.clawperator.com.
For agents
Prefer the technical docs over this overview page. Start with the Node API guide, the Operator LLM playbook, or the machine-oriented markdown landing page.
Built for agent loops
More than UI automation. Each command does one thing, returns one result, and never hides retries.
action→execution→structured result
Example agent loop (pseudocode - see docs.clawperator.com for the full API)
open_app(<google_home_app_id>)
snapshot_ui()
click(<climate_tab>)
snapshot_ui()
scroll_and_click(<device_labeled_"Living room">)
snapshot_ui()
if <hvac_state> == "Off":
click(<turn_on>)Deterministic
Each command does one thing, returns one result, and never hides retries.
Structured
Agents get machine-readable UI state, explicit errors, and results they can branch on.
Built for loops
Clawperator is designed for reasoning systems that need predictable execution, not best-effort automation.
Skills
Clawperator automations are unlocked through skills.
Clawperator includes an open source, ever-expanding skills repository for common Android workflows.
But you are not blocked waiting for an official skill. Agents do not need a prebuilt skill to automate your apps.
Our dedicated build-your-own-skill-from-scratch documentation walks agents through the process step by step. Point your agent at the right app and tell it to make a skill. It can use the clawperator API to inspect the app's UI, find a reliable path to the state or action you need, and create a private, personalized skill for your exact workflow.
Included with install
- Open source skills repository
- Reusable building blocks for common workflows
- Discoverable and runnable through the same runtime
Yours to create
- Private skills for your own apps and accounts
- Agent-built skills from live app exploration
- Public and private skills mixed in the same runtime
You are not blocked waiting for a public skill. Start with included skills, adapt them to your setup, or let your agent create private ones through Clawperator's documented API.
How it actually works
Your agent is the brain. Clawperator is the hand. Skills sit above the runtime as reusable app-specific workflows. Whether the agent runs a skill or drives the UI step by step, Clawperator is the execution layer that talks to the Android device and returns structured results.
AI Agent / LLM
Node API / CLI / Skills
Clawperator
ADB
Android Device
Mobile Apps
The runtime includes the CLI on your host machine and the Clawperator Operator Android app on the Android device. The agent decides what to do next. Skills give the agent reusable ways to handle app-specific workflows it already understands, whether those come from the open source skills repository or from private skills the agent created for you. Clawperator executes the Android side of the workflow and returns data your agent can use.
When a skill exists
- The agent picks a skill for a known app workflow
- The skill packages the reliable path to the state or action needed
- Clawperator still executes the underlying Android actions and returns the result
When no skill exists yet
- The agent opens the app and inspects the live UI
- It finds a reliable path step by step using the documented API
- Once the flow is understood, it can turn that path into a private skill
The operating loop is simple: observe, decide, execute, return.
01
Observe
Capture the current Android UI as structured state so the agent can reason on a real snapshot of the app.
02
Decide
The agent chooses the next action or runs a skill. Clawperator does not plan, improvise, or decide on its own.
03
Execute
Clawperator performs the requested tap, type, scroll, read, or app action on the connected Android device.
04
Return
Return one machine-readable result so the agent can continue, recover, or stop with confidence.
FAQ
Can I use this if I have an iPhone?
Yes. Clawperator does not require you to switch your primary phone to Android. The normal setup is to keep using your iPhone and connect your agent to a separate Android burner phone or a local Android emulator.
I do not have an Android phone. Can I still use Clawperator?
Yes. Clawperator can provision a local Google Play-equipped Android emulator. That is useful for getting started, local development, and many automation flows.
Do I need a dedicated burner phone?
A cheap dedicated burner phone is the preferred setup for compatibility and long-running reliability, but it is not the only option. A local Android emulator is available when you do not have a device handy.
What if the skill I need does not exist yet?
That is fine. Clawperator includes a growing open source skills library, but you are not blocked on prebuilt skills. Your agent can use Clawperator's structured, documented API to explore an app, operate it, and build a private skill for your own workflow.
Does Clawperator do the thinking for my agent?
No. The agent decides what to do next. Clawperator executes validated Android actions and returns structured results.