- Home
- Skills
- APIs & Backend
- hardware controller
More screenshots
Works with the AI Agents you already use
hardware controller
Control and program robots using SQL hardware declarations, Python/Lua drivers, and MT action scripts.
Free
hardware controller
Example session with this skill installed
Now scan I2C-1
- Read your context and instructions
- Compiled the hardware controller
Scan worked this time — the device node is now crw-rw-rw-, so permissions are fixed.
/dev/i2c-1 result
{"ok":true,"outputs":[{
"result":"OK","bus":"/dev/i2c-1",
"found":["0x40","0x41"],"count":2,"range":"0x08-0x77"
}]}
Two devices ACK: 0x40 and 0x41.
Adapter behind that bus: Synopsys DesignWare I2C adapter — the Intel PCH SMBus controller, not a GPU/DDC line. Adjacent
addresses like that usually mean two instances of the same chip: common candidates are INA219/INA226 power monitors,
PCA9685 PWM drivers, or a pair of temperature sensors. Can't name them for certain without reading their registers.
Want me to sweep i2c-0 and i2c-2 through i2c-10 too? That would map what's actually on the board, and I can note which
addresses respond on more than one bus.
Connects securely to your tools. The creator never sees your data.
About this skill
Programming hardware requires juggling low-level drivers, timing-sensitive sequences, and complex wiring declarations. This skill provides a self-contained Python runtime that separates hardware structure, driver implementation, and action sequencing into a manageable three-layer model.
What it does
- Hardware Declaration uses Structured SQL to define devices, buses (I2C/SPI/UART), and address mapping.
- Driver Implementation supports both Python and Lua for writing device logic, lifecycle methods, and HAL calls — or select from 40+ built-in drivers.
- Sequence Orchestration uses MT action scripts to describe time-series actions and nested logic blocks.
- Static Validation checks references, driver existence, and syntax without activating physical hardware.
- Hardware Simulation enables full-link verification using a virtual protocol and built-in virt-device models.
- Real-time Sampling reads sensor data, daemon snapshots, and variable states directly from the CLI.
How it works
- Environment Check run the
doctorcommand to verify dependencies, Lua runtimes, and asset directories. - Declaration write a
.sqlfile to define the robot structure — most common devices already have built-in drivers, or write a.py/.luadriver for custom hardware. - Verification execute the
checkcommand to catch syntax and reference errors, thenrunwithproto='virt'for full-link simulation. - Execution deploy to hardware using the
runcommand with the--allow-hardwareflag for physical I/O.
Frameworks & tools
Built on a custom Python runtime (pymybot) with support for Lua scripting. Exposes controller tools to MCP clients for external agent integration, and provides a stdio JSON-RPC backend for GUI/IDE connectivity.
Why this beats prompting it yourself
Prompting an LLM to generate hardware timing scripts often results in syntax hallucinations and dangerous I/O operations. This skill enforces a strict schema, provides a dedicated validation layer, and ensures hardware is only accessed through controlled, explicit permissions.
Use cases
- Prototyping developing I2C/SPI sensor drivers using virtual models before physical hardware arrives.
- Automated Testing running time-series hardware checks and motion sequences for quality assurance.
- Remote Orchestration exposing robot controls to an LLM via the built-in MCP server for high-level tasks.
- Field Debugging sampling live sensor values and daemon logs to troubleshoot wiring or driver failures.
Known limitations
Does not include camera/vision tools, which are handled by native host agents. Physical hardware access and non-loopback networking require explicit manual flags for security.
Requires manual flags for physical hardware access. No built-in vision/camera support. Commands output single-line JSON. Driver failures in dispatch state may result in unknown outcome status.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 11 days ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts