Menú
describes how to write code to be read and validated by human scientists.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Use this skill for quantitative scientific analysis of IBL data after the always-on scientific workflow in `AGENTS.md` is active.
Use this skill when loading IBL data, preferring the smallest sufficient local or aggregate surface before session-level loaders.
Use this skill for raw Neuropixels binary access, destriping, compression, waveform extraction, and probe-level preprocessing based on the `int-brain-lab/ibl-neuropixel` repository.
Use this skill to analyze scientific data: refine, formalize and test hypotheses.
Use this skill if the required tools are not yet installed.
Use this skill when a task needs IBL data access setup, ONE/Alyx authentication, or selecting between public and private data modes.
| name | scientific-coding-style |
| description | describes how to write code to be read and validated by human scientists. |
The code you write will be read and checked by scientists, not software engineers.
Keep it simple and short. The user will read all of the code so brevity is essential.
Avoid defensive programming, catching edge cases (try/catch) and type checking unless absolutely necessary. It is better for code to occasionally crash than to be complex. It is better to crash than to silently produce incorrect results.
Avoid classes unless they increase clarity. Avoid clever code that is hard to read.
Do use vectorization wherever possible. For speed and also brevity; scientists will find this easier to read than loops.
Use comments to explain the scientific reasoning behind your code. Use descriptive variable names. Each function should have a docstring explaining what it does, what its inputs and outputs are (including array sizes), and any assumptions it makes.
