add-step

name	add-step
description	Scaffold a new pipeline step. Creates script file, updates pipeline.md, 00_run.do, and run_all.sh. Use when adding new analysis steps.
allowed-tools	["Read","Write","Edit","Bash","Glob"]

Add Pipeline Step

Scaffold a new pipeline step with all necessary documentation and integration.

Steps

1. Gather information (ask user if not provided):

   • Step number (suggest the next available number with a gap)
   • Script name and language (Stata, Python, or SAS)
   • Brief description of what the step does
   • Input files
   • Expected output files

2. Create the script file with a standard header:

For Stata (code/stata/NN_description.do):

* ==============================================================================
* [Step NN]: [Descriptive Title]
*
* Purpose: [What this script does]
* Inputs:  [Data files read]
* Outputs: [Files created]
* Dependencies: [Upstream scripts]
* ==============================================================================

clear all
set more off

* Load globals from master do-file
do "$stata/00_run.do"

* --- Main code here ---

For Python (code/python/NN_description.py):

"""
[Step NN]: [Descriptive Title]

Purpose: [What this script does]
Inputs:  [Data files read]
Outputs: [Files created]
Dependencies: [Upstream scripts]
"""

from pathlib import Path
import pandas as pd

# --- Main code here ---

For SAS (code/sas/NN_description.sas):

/* ==============================================================================
 * [Step NN]: [Descriptive Title]
 *
 * Purpose: [What this script does]
 * Inputs:  [Data files/libraries read]
 * Outputs: [Datasets/files created]
 * Dependencies: [Upstream scripts, WRDS access needed?]
 * ============================================================================== */

%let projroot = [PROJECT ROOT PATH];
libname raw    "&projroot/data/raw";
libname proc   "&projroot/data/processed";

/* --- Main code here --- */

3. Update pipeline.md:

   • Add the step to the Pipeline Tree in the correct position
   • Add to the Script Status table
   • Add to the Data Files table if it creates new data
   • Add to the Manuscript Figure Manifest if it creates figures/tables

4. Update code/stata/00_run.do:

   • Add a commented-out do line in the correct position

5. Update run_all.sh:

   • Add a commented-out run_script line in the --all section

6. Show the user everything before writing.

Important

• This changes the pipeline -- "break the glass" rule applies
• Show all proposed changes to the user before writing
• Use step number gaps (01, 05, 10, 15, 20, 25) for easy future insertion