Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Neuropixels Analysis (SpikeInterface)

End-to-end Neuropixels extracellular analysis with SpikeInterface: loading, preprocessing, drift correction, spike sorting, quality metrics and unit curation.

Data & AnalyticsAdvanced★ 33,030⑂ 3,248AI score 8/10Last updated: Aug 9, 2026

What it does

Guides Claude through a complete Neuropixels 1.0/2.0 analysis workflow built on SpikeInterface.

  • Loading: SpikeGLX (.ap.bin/.meta), Open Ephys, NWB, with stream inspection first
  • Preprocessing: highpass filter, bad-channel detection/removal, phase_shift ADC correction, global common median reference
  • Drift check & correction: peak detection/localization, drift raster plots, correct_motion presets (dredge, nonrigid_fast_and_accurate, ...)
  • Spike sorting: Kilosort4 (GPU), CPU sorters (SpykingCircus2, Tridesclous2, Mountainsort5), containerized external sorters
  • Postprocessing & metrics: SortingAnalyzer, waveforms/templates/correlograms, SNR, ISI violations, presence ratio, amplitude cutoff
  • Curation: Allen/IBL threshold queries, Hugging Face UnitRefine model classifiers (noise vs neural, SUA vs MUA), and agent-based visual review of borderline units
  • Export: Phy, PNG report, NWB, CSV metrics

Runnable scripts in scripts/, a copy-and-edit template in assets/, and topic reference guides ship with the skill.

Who it's for

  • Neuroscientists and grad students running in vivo electrophysiology with Neuropixels probes
  • Labs standardizing a reproducible spike-sorting pipeline
  • Anyone wanting to reduce manual Phy curation with automated/model-based triage
  • Newcomers to the SpikeInterface API who want Allen/IBL-style best practices

Example uses

  1. Sanity-check a new recording: "List the streams, channel count, duration and bad channels in this SpikeGLX folder" → run explore_recording.py, then slice the first 60 s for a fast trial run.
  2. Run the whole pipeline: python scripts/neuropixels_pipeline.py /data output/ --sorter kilosort4 --curation allen, then ask Claude to interpret the resulting quality-metrics table.
  3. Adjudicate borderline units: render waveform/correlogram summaries for ambiguous units, have Claude comment on isolation quality, and export the survivors to Phy for a final human pass.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/neuropixels-analysis folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/neuropixels-analysis/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/K-Dense-AI/scientific-agent-skills.git && mkdir -p ~/.claude/skills && cp -r scientific-agent-skills/skills/neuropixels-analysis ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Confirm Python ≥ 3.10 (python --version). If you plan to use Kilosort4, you also need a CUDA GPU with drivers installed.
  2. Clone the skill repository: 
    git clone https://github.com/K-Dense-AI/scientific-agent-skills.git
  3. Copy the skill folder into your Claude Code skills directory: 
    mkdir -p ~/.claude/skills
cp -r scientific-agent-skills/skills/neuropixels-analysis ~/.claude/skills/
  4. Create a Python environment and install the core stack: 
    uv pip install "spikeinterface[full]" probeinterface neo
  5. Install the sorter you need: uv pip install kilosort (GPU) or uv pip install mountainsort5 (CPU).
  6. For model-based curation add uv pip install huggingface_hub skops; for AI-assisted visual curation add uv pip install anthropic.
  7. (Optional) Export ANTHROPIC_API_KEY in your shell for programmatic vision calls — never hardcode keys in analysis scripts.
  8. Restart Claude Code and ask something like "analyze this Neuropixels recording" to trigger the skill.
View source on GitHub ↗License: MIT