Skip to content

Skill: pipeio-flow-create — guided flow creation workflow

Problem: Creating a new pipeline flow requires assembling ~10 conventions from CLAUDE.md, pipeio-guide, and existing flows. During the pixecog preprocess_motion creation session, an agent hit multiple frictions: overwrote scaffolded files, removed configfile directive incorrectly, forgot to wire conditional rules, didn't know about registry_scan, etc.

Proposed skill: pipeio-flow-create

A skill that guides agents through the full flow creation workflow:

  1. Survey phase — examine existing flows for patterns (Snakefile structure, config schema, script template)
  2. Scaffold — call pipeio_flow_new, then pipeio_registry_scan
  3. Config — write config.yml following the established schema: pybids_inputs, registry groups (with base_input, bids root/datatype, members), processing params
  4. Snakefile — write rules following the pattern: configfile: "config.yml" + manual safe_load override, BidsPaths adapter, in_paths/out_paths, dynamic input functions for conditional rules
  5. Scripts — write processing scripts with the standard template: docstring, snakemake mock, repo_abs, mkdir for outputs
  6. Datalad — create subdataset: datalad run-procedure create-reckless-subdataset derivatives/<name>
  7. Verify — dry-run via pipeio_run(dryrun=True)

Key conventions to encode: - Don't overwrite scaffolded files (Makefile, notebook.yml) — only customize what's needed - configfile: "config.yml" is correct — flows run from their own directory - Use pipeio_run MCP tool, not manual conda run / snakemake commands - Pipeline env is cogpy - After scaffold, must call pipeio_registry_scan - Registry groups need base_input matching a pybids_inputs key - Scripts use from sutil.repo_root import repo_abs for path resolution

Source context: pixecog

PixEcog (pixecog): Neuropixels and ECoG dataset and analysis

Recent commits: 

27a75e5 Preprocess_ieeg DAG redesign: ttl_removal before interpolate, two-tier badlabel, auto-skip TTL, pre-lowpass LNR
42cf0d9 Preprocess_ieeg redesign: DAG reorder, two-tier badlabel, auto-skip TTL, pre-lowpass LNR, cross-subject TTL analysis, notebooks published
6b295b2 Update badlabel audit note: full pipeline comparison (85 new vs 7 legacy vs 71 TTL), zarr fix confirmed, 97.2% TTL catch rate

README:

type: readme

Quick Start for Collaborators

Follow this checklist to get started with Pixecog documentation and workflows.

🐀 Pixecog Project — Compact Overview

Core principles

  • One immutable BIDS raw dataset (raw/) as the canonical baseline
  • Each analysis pipeline ha