survey-context
Skill2 starsChanged 3 months agoMITView on GitHub
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].
1 other project uses a version of this.
Adopted by
Most recently taken up by
- danielvm-git/bigpowerstheir own version, first seen Nov 6, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- danielvm-git/bigpowers★ 292same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- data-checkSkill
- daten-checkSkill
- decision-briefSkill
- entscheidungsvorlageSkill
- executive-summarySkill
- freitext-themenSkill
- hypothese-checkSkill
- hypothesis-checkSkill
- management-summarySkill
- segment-profileSkill
- segment-profilSkill
- survey-reportSkill
- umfrage-kontextSkill
- umfrage-reportSkill
- verbatim-themesSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 46
Run from the project root.
- Very specific: line 57
Check for an existing document: If <name>.context.md already exists, read it, summarize in two lines what it already covers, and continue …
- Distinctive: line 63
Was a briefing supplied?: If the user's request (or a file they point at) already contains project background, take everything you can from …
- Distinctive: line 80
Ask them in this order, in the user's own words, one message each.
- Distinctive: line 93
Fieldwork: When did the survey run, and did anything happen shortly before or during it that shapes the answers — a release, a price …
- Very specific: line 147
Every section stays in the document even when unanswered.
- Very specific: line 160
Do not answer analysis questions here: If the user asks what the data show, point them at survey-report and finish the interview first.
- Distinctive: line 162
Document language = language the user speaks (default English).
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
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.mdThe 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
- Find the CSV. A path in the user's request wins; otherwise the CSV in the current folder. With several CSVs and no path given, ask which one — the document is bound to a specific data set.
- Check for an existing document. If
<name>.context.mdalready exists, read it, summarize in two lines what it already covers, and continue at the first section that is empty or marked as not stated. Never silently overwrite answers that are already there; ask before changing an existing section. - Run
profile. Note n, the number of columns and the question wordings — they tell you a lot about the subject already. - Was a briefing supplied? If the user's request (or a file they point at) already contains project background, take everything you can from it, write the document, and then only ask about the sections that stayed empty. The interview fills the format; it is not the only way to produce it.
- Create the document immediately, with the header and empty sections, before asking anything. Then ask the six questions one at a time, and update the file after each answer. An interview abandoned after question three must leave a usable document behind.
- Ask the six questions (below). Say up front that it is six questions, roughly two
minutes, and that anything can be skipped. Show progress ("Question 3 of 6"). If an
answer is skipped, write
Not stated.into that section — an explicit gap is information; silence is indistinguishable from "there is no limitation here". - Finish: give the path, note that the file is gitignored, and say in one sentence that the other skills will use it from now on.
The six questions
Ask them in this order, in the user's own words, one message each. The parenthetical notes are for you, not to be read out.
- Purpose. Why did this survey run, and which decision depends on it?
- Subject. What is the product or topic in a sentence or two — and which terms or
abbreviations in the column names would an outsider not understand?
(You have the column wordings from
profile; ask specifically about the ones that look like internal shorthand.) - Recruiting. Whom did you want to survey, and how did participants get in? (Then follow up once: who is systematically missing as a result? This is the single most valuable sentence in the document. If the user does not see a gap, offer what the recruiting route implies — an in-app banner cannot reach churned users, a customer newsletter cannot reach non-customers — and let them confirm or correct it.)
- Fieldwork. When did the survey run, and did anything happen shortly before or during it that shapes the answers — a release, a price change, an outage, press coverage?
- Audience. Who reads the reports, and what should they do with them?
- Expectations. What do you expect the data to show? (Say plainly what happens with the answer: it goes into the document as a list of things to test, and a report that refutes an expectation is doing its job. Do not press if the user has none — an empty expectations section is fine.)
Document format
# Survey context: <short title>
**Data set:** <file.csv>
**Sample:** n = <responses> · <number> columns
**Fieldwork:** <period, as stated by the user>
**Context recorded:** <date>
## Purpose and pending decision
Why the survey ran and what is to be decided from it.
## Subject and glossary
What the product/topic is. Terms, abbreviations and internal names that the column
wordings do not explain.
## Sample and recruiting
Target population and how participants were recruited.
## Who is missing
Which groups the recruiting route cannot reach, and what that means for the findings.
This section belongs in the limitations of every report.
## Circumstances during fieldwork
Events during or shortly before fieldwork that may have shaped the answers.
## Column notes
Which column is the lead question, which items belong together, which columns carry the
segments the business cares about. Free prose, no fixed notation.
## Audience and use
Who reads the reports and what they decide with them.
## Expectations (to be tested)
**These are prior expectations of the people who ran the survey. They are something to
test, never something to support.** Do not cite an expectation as evidence, do not let it
influence which analyses are run or how a finding is worded. Report an expectation only
where the data actually answer it, and then with an explicit verdict — including "the data
do not show this".
- <expectation 1>
- <expectation 2>Every section stays in the document even when unanswered; write Not stated. underneath.
Rules
- Record, do not invent. Everything in the document comes from the user or from
profile. If an answer is vague, ask once for a concrete detail, then write down what was actually said — not a tidied-up version of it. - Keep facts and expectations apart. Nothing from question 6 may end up in any other section. If an answer mixes them ("we surveyed our power users, and they are surely happier"), split it and say that you did.
- Short beats complete. Two to four sentences per section. This is a briefing note, not a study protocol.
- No personal raw data — no respondent names, ids or identifying quotes.
- Do not answer analysis questions here. If the user asks what the data show, point
them at
survey-reportand finish the interview first. - Document language = language the user speaks (default English).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

