agentleFS
lenar-amirovlenar-amirov/product-pipeline-public/.claude/skills/experiment-design/SKILL.md

experiment-design

Skill13 starsChanged 5 months agoMITView on GitHub

Designs the validation experiment for a solution's riskiest assumption — AB test parameters (MDE, sample size, duration, guardrails, decision criteria) or a cheaper pre-test. The engine of /design-ab-test (step 14). Use for "спланируй AB-тест", "design the experiment", "sample size", "MDE", "как проверим решение", "критерии успеха теста".

32 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use list-quality-scorecard and positive-reply-scoring.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • creatop-gtmSkill

    list-quality-scorecard

    Check a lead list before it goes into a sequence. Gives a deliverability readout (firm) and a targeting readout (advisory). Surfaces issues, never blocks. Use before enrolling any list.

    Also in 3 projects that use this

  • creatop-gtmSkill

    positive-reply-scoring

    Classify replies and compute positive reply rate, the metric that predicts revenue. Use to tell whether a campaign is actually working, not just getting replies.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 23 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

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

    Verify the events the test needs actually exist (tech-spec §tracking).

    line 44

“Common” is measured: kept by at least 6 in 10 of the 23 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

Experiment Design — /design-ab-test engine

Input: the top solution from solution-scoring and its riskiest assumption (from the assumption map). The experiment tests THAT, not "the feature in general".

0. Is an AB test even the right instrument?

Cheaper first: a concept test (user-testing skill) kills desirability risks in days; a fake-door sizes demand; an ops pilot tests feasibility. AB test is for causal measurement of the metric effect at scale — expensive and slow; don't default to it. State the choice and why.

1. AB test parameters (with the analyst)

  • Hypothesis in the falsifiable form: "If [change] for [segment], then [primary metric] moves from [baseline] by ≥[MDE], because [mechanism]." Baseline comes from the registry/CONTEXT.md, never from memory.
  • Primary metric — exactly one; the same definition the dashboard uses (metric-definition mismatches are how gates die — see /challenge).
  • MDE — the smallest business-meaningful effect (the shipping floor), NOT the smallest statistically detectable one and not the effect you hope for. A result can be significant and still worthless — sizing the test to detect 0.2% when only 2%+ changes the roadmap wastes weeks.
  • Sample size & duration — from baseline, MDE, power 0.8, α 0.05; respect weekly seasonality (full weeks only). If the segment is too small for the MDE — say the test is infeasible and go back to instrument choice.
  • Guardrails — metrics that must NOT degrade (engagement, latency, support load) with explicit thresholds.
  • Decision criteria are pre-registered: ship / iterate / kill thresholds written into this doc BEFORE launch — post-hoc goalpost moving is how p-hacking enters product work. Division of power: the primary metric decides ship/no-ship; a guardrail breach pauses the test and triggers investigation — it neither silently vetoes nor gets explained away after the fact.

2. Tracking readiness

Verify the events the test needs actually exist (tech-spec §tracking). A test on unmeasured behavior is a dependency, not a test — create it in status.json with the analyst as owner.

3. Output

output/ab-test-design.md: hypothesis, parameters table, guardrails, decision criteria, experiment IDs placeholder. Create the AB-test dependency (owner: analyst, deadline: launch + duration). Step 16 (/analyze-ab-test) reads this file as its contract — vague criteria here become unfalsifiable results there.

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.