embedino GitHub

Open project · Early development

AI can write code.
Give it the hardware context.

Embedino is building a harness for AI-assisted firmware development. It brings board details, connected components, pin mappings, constraints, and build feedback into one workflow so firmware can be written for the hardware you actually have.

Explore the project See the approach

The harness is in development. The diagram shows the intended workflow.

01 / THE PROBLEM

Firmware depends on facts a prompt often misses.

A generated program can look correct and still target the wrong pins, voltage, peripheral, board variant, or library. Those details usually live in datasheets, wiring notes, build settings, and test logs.

Embedino’s goal is to make that context explicit and reusable throughout the firmware workflow.

02 / THE APPROACH

A harness around the model, built for real boards.

We’re designing a loop that connects hardware description, code generation, and engineering feedback.

01

Describe the hardware

Record the board, components, pin assignments, interfaces, and constraints in a format the AI can use.

02

Write with context

Give the assistant the relevant hardware facts and project requirements when it drafts or changes firmware.

03

Learn from the build

Bring compiler output and on-device observations back into the loop so the next revision addresses real failures.

03 / PROJECT STATUS

Building the foundation in the open.

The public repository currently explains the product direction. The first harness modules and usable release are still being developed. We plan to integrate Claude for firmware drafting and revision; that integration is not live yet.

If you build embedded systems, your feedback on the workflow and failure cases is welcome.

Follow progress on GitHub