survey-contextlisted

Captures the context of a survey — purpose, subject, recruiting, who is missing, audience and expectations — in a short interview and stores it as a Markdown document next to the CSV (<name>.context.md). Every other survey skill reads it when present, so reports state the study design instead of guessing it. Use before the first analysis of a new data set, to update an existing context document, or to turn an existing briefing into one. Invoke with your CSV path or briefing text as the argument, e.g. /survey-context [CSV path].
kayspiegel/survey-insight-skills · ★ 2 · Data & Documents · score 76

Install: claude install-skill kayspiegel/survey-insight-skills

# Survey Context Goal: record **what the data cannot say about itself** — why the survey ran, whom it asked, how people were recruited, who is therefore missing, and what decision is pending. The result is one short Markdown document next to the CSV. All other skills in this toolkit read it when it exists, so their reports describe the actual study instead of a plausible generic one. This is an **interview**, not a form. Keep it short and never guess an answer on the user's behalf. ## Where the document lives Next to the CSV it describes, named after it: ``` kundenumfrage.csv → kundenumfrage.context.md ``` The suffix is `.context.md` in every language — only the contents are written in the user's language. If the folder holds exactly one CSV, a plain `survey-context.md` is accepted too (that is the name to use when writing one by hand without a data file yet). The file is ignored by `.gitignore` on purpose: it can be more sensitive than the CSV, because it names pending decisions, strategy and internal metrics. Mention this once at the end. ## Tool ``` python3 scripts/survey.py profile [--file CSV] ``` Run from the project root. If `scripts/survey.py` is not there, check `.claude/scripts/survey.py` or locate `survey.py` in the project. `profile` gives n, the column list with indices and types, and fill rates. Use it so the interview never asks for anything the data already state. ## Workflow 1. **Find the CSV.** A path in the user's request wins; otherwise the