Pipeline for Neuropixels extracellular electrophysiology: probe geometry (ProbeInterface), Kilosort sorting via SpikeInterface, quality metrics, unit curation (ISI, firing rate, SNR), post-sort analysis (PSTH, tuning curves, population decoding). Supports Neuropixels 1.0/2.0/Ultra in rodent/primate experiments.
SkillScore86Maintained
Automatic
◆Research pack
Last verified 8/24/2026
What it changes
Passed 18 of 21 test cases with the skill, and 14 of 21 without it.
Without skill68%
With skill86%
Checks it passed that the model alone missed
The sorter call sets nblocks=5.
The response recommends lowering Th_learned to 6.
The maximum allowable amplitude cutoff threshold is 0.1.
and 1 more in the full report
4 fixed15 passed either way3 still failing0 broke
Pass-rate lift
+18 pp
Tokens
+186%
Agent turns
0%
Response time
-17%
Model gemini-3.6-flashMeasured 8/24/2026v119 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 3 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 3
Prompt
After running Kilosort4 on a filtered Neuropixels recording, the sorter returned 0 units. A colleague guessed that the probe mapping was deleted. Based on standard Kilosort4 troubleshooting, how should you adjust `Th_learned` to resolve zero units found?
What was checked, without and with the skill
Checked
Without skill
With skill
The response recommends lowering Th_learned to 6.
Missed
Passed
case-11 · gemini-3.6-flash · 8/24/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md330 linessingle file
Neuropixels Analysis
Overview
Neuropixels probes record extracellular voltage from 384 (NP 1.0) or 192 (NP 2.0) simultaneously recorded channels at 30 kHz. Analysis follows a canonical pipeline: raw data → spike sorting (Kilosort) → quality curation → unit analysis. SpikeInterface (Python) provides a unified API across 10+ spike sorters, handles data loading from multiple formats (SpikeGLX, OpenEphys, NWB), computes quality metrics, and exports sorted results. ProbeInterface manages probe geometry and channel maps. Post-sort analysis (PSTHs, firing rate, decoding) uses standard Python scientific stack.
When to Use
Spike-sorting Neuropixels recordings from SpikeGLX (.bin) or OpenEphys (.dat) to extract single-unit activity
Applying automatic quality control metrics (ISI violations, SNR, firing rate) to curate sorted units
Computing peristimulus time histograms (PSTHs) locked to experimental events
Analyzing population coding: decoding stimulus or behavioral variables from firing rates
Converting sorted data to NWB (Neurodata Without Borders) format for sharing
Comparing multiple spike sorters on the same dataset for method validation
Visualizing unit waveforms, auto-correlograms, and spatial distribution across probe channels
Use SpikeInterface instead for a unified framework that supports 10+ spike sorters with a common API and comparison tools
import spikeinterface.full as si
from pathlib import Path
# SpikeGLX recording (most common Neuropixels format)
data_dir = Path("/data/recording/session_001")