pudap/bears-skills
2 skills2 installsGitHub
Install
npx skills add https://github.com/pudap/bears-skillsSkills in this repo
1Bears MachinesBears Machines (balance-machine) is an agent skill for builders wiring a Linux-hosted Arduino mass balance into automated experiments. It documents how raw serial ADC samples become calibrated grams, how the edge service connects to NATS for telemetry, and how to configure serial ports, baud rates, and environment files before starting an unattended edge process. Solo indie lab or hardware-adjacent developers use it when liquid-handling or gravimetric workflows need live weight feedback instead of manual scale reads. Prerequisites center on Python 3.11+, uv, a device on /dev/ttyUSB* or /dev/ttyACM*, and a reachable NATS server. The skill is narrow and integration-focused: it does not replace mechanical calibration science, but it gives agents a concrete path from `.env` through `balance.py` to streaming mass events.1installs2Bears WorkflowsBears Workflows (documented in-repo as colour-mixing optimization) is a lab-automation agent skill for solo builders and small teams running closed-loop experiments on an Opentrons OT-2 with an attached camera. It enforces a sequential single-run discipline: one active run at a time, a fresh run_id each iteration, and gated progression until the run is verified successful. Before any command generation, the agent must invoke companion skills for machine access, protocol authoring, persistent experiment logs, and report output paths. Optimization can follow efficient Bayesian Optimization over volume ratios or an LLM approach when constraints need flexible reasoning. The workflow targets iterative RGB mixing with RMSE against a target colour using live imaging feedback. It is highly specialized hardware software—not a general CI workflow—and assumes OT-2 plus camera hardware and the broader puda skill stack.1installs