agentleFS
kayspiegelkayspiegel/survey-insight-skills/.agents/skills/decision-brief/SKILL.md

decision-brief

Skill2 starsChanged 3 months agoMITView on GitHub

Builds a structured decision brief (decision memo) from quantitative survey data (CSV): situation, options with pros/cons backed by data, a reasoned recommendation, risks and assumptions. Result as Markdown to reports/. Use when a decision is due that should be grounded in research evidence. Invoke with your request as the argument, e.g. /decision-brief <decision / question>.

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run these from the project root.

    line 24
  2. Very specific:

    If a context document sits next to the CSV (<name>.context.md, or survey-context.md when the folder holds exactly one CSV), read it first — …

    line 34
  3. Distinctive:

    If not given, derive 2–4 realistic options (include "status quo / do nothing" as a baseline).

    line 47
  4. Distinctive:

    Use significance/effect size where group differences drive the decision.

    line 95

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Decision Brief (Decision Memo)

Goal: For the decision named in the user's request, build an actionable brief in which every option is backed by data and a clear recommendation is given. For decision-makers, not analysts. Result as Markdown in reports/.

Tool

python3 scripts/survey.py <cmd> [...] [--file CSV]

Run these from the project root. If scripts/survey.py is not there, check .claude/scripts/survey.py or locate survey.py in the project.

  • profile — column overview (always first).
  • freq COL [--by COL2] [--filter "COL=VALUE"] [--sig] — distribution/cross-tab, --sig for significance + effect size (single × single).
  • text COL [--filter ...] [--sample N] — verbatims/free text as evidence.

COL = index (from profile) or an unambiguous name part.

Survey context

If a context document sits next to the CSV (<name>.context.md, or survey-context.md when the folder holds exactly one CSV), read it first — this skill benefits from it more than any other, because it names the pending decision and the audience the brief is for. Take the situation, the subject and the glossary from it; the figures still come only from the engine. Its recruiting route and "Who is missing" section belong under risks and assumptions: an option that looks strong among the people who answered may look different among those the survey never reached. Expectations listed there are something to test, never something to support — never let one decide which option wins. If no such document exists, proceed as before and mention survey-context once at the end.

Approach

  1. Sharpen the decision. What exactly is being decided? Which options are plausible? If not given, derive 2–4 realistic options (include "status quo / do nothing" as a baseline).
  2. Profile and find the decision-relevant variables.
  3. Gather evidence per option. For each option, query the data that argues for or against it (distributions, cross-tabs, significance where relevant, free-text needs). Also quantify the size of affected user groups (impact).
  4. Weigh and recommend. Compare options on the evidence, give a reasoned recommendation. Expose trade-offs, don't hide them.
  5. Save the report to reports/ (date +%F + kebab title, never overwrite) and state the recommendation + path in your reply.

Report format

# Decision brief: <decision>

**Data basis:** <file> · n = <…> · <date>

## Recommendation
Clear statement in 1–2 sentences: which option, and why.

## Situation
Why is the decision due? Relevant data context in 2–4 sentences.

## Options compared
| Option | For (data) | Against (data) | Affected (impact) |
|---|---|---|---|
| A … | … (number) | … (number) | … % of users |
| B … | … | … | … |
| Status quo | … | … | … |

## Rationale
Why the recommended option wins the trade-off — with the decisive numbers.

## Risks & assumptions
What must hold? What could go wrong? Data gaps/uncertainties, including whom the survey
could not reach (from the survey context) and what that could mean for the recommendation.

## Next steps
2–4 concrete, prioritized steps.

## Method
Columns/filters/tests used, in one paragraph.

Rules

  • Every option needs data (pro and con) — no gut calls.
  • State the recommendation unambiguously, but make trade-offs and uncertainty transparent.
  • Use significance/effect size where group differences drive the decision.
  • Quantify impact (how many users / what share affected).
  • Report language = language of the request (default English). Decision-maker tone: short, concrete.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.