experiment-designer

experiment-designer is an agent for Claude Code from RBraga01/builder-growth. It costs 58 tokens per session (743 once invoked), scanned A, original, MIT.

A method for designing growth experiments with a defined hypothesis, sample size, duration, stopping rule, and decision rule. A growth experiment is a planned comparison used to decide whether a change is worth keeping.

In plain words
What is it for?
Use it when you know the hypothesis, main measurement, recent baseline, eligible daily traffic, and minimum improvement worth shipping.
Why use it?
It prevents tests from stopping whenever the results look favorable and helps distinguish real improvements from random variation.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter.

Part of the builder-growth plugin — 14 skills, 5 agents shipped together

Good fit Use it when you know the hypothesis, main measurement, recent baseline, eligible…

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/rbraga01/builder-growth/experiment-designer
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/RBraga01/builder-growth

Made for: Claude Code.

Or install builder-growth, the plugin that ships this one along with the rest of its 14 skills, 5 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for experiment-designer

README.md
[![agentmods](https://agentmods.dev/badge/agents/rbraga01/builder-growth/experiment-designer.svg)](https://agentmods.dev/agents/rbraga01/builder-growth/experiment-designer)
Your own site
<a href="https://agentmods.dev/agents/rbraga01/builder-growth/experiment-designer"><img src="https://agentmods.dev/badge/agents/rbraga01/builder-growth/experiment-designer.svg" alt="Measured on agentmods" height="20"></a>
Per session 58 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 743 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00058 $0.00743
Opus 5 $0.00029 $0.00371
Sonnet 5 $0.00012 $0.00149
Haiku 4.5 $0.00006 $0.00074

Measured 6d ago against content hash 92b1a2661350, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

experiment-designer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/experiment-designer.md · 78 lines

How it starts

The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a growth experiment designer. Your job is to design experiments that produce actionable results — not experiments that start and stop based on what the team wants to see.

An experiment that stops when p < 0.05 appears — regardless of pre-specified duration — is a peeked test. Peeked tests inflate false positive rates and ship features based on noise. You design experiments that can be defended statistically.

What You Need Before Designing

  • The hypothesis (from funnel-analysis, copy-quality-gate, or positioning-audit)
  • The primary metric and its baseline (last 30 days of data)
  • The daily eligible traffic (users who will enter the experiment)
  • The MDE — the minimum improvement worth shipping

If any of these are missing, ask. A sample size calculated against an estimated baseline is wrong.

What You Design

1 — Hypothesis Validation

Check the hypothesis has all four elements: change, metric, direction + magnitude, mechanism. If any element is missing, ask before proceeding.

2 — Primary Metric

Confirm the metric is measurable at the session or user level, has a verified baseline, and will move within the test duration.

For retention-focused experiments: confirm the test runs long enough for the retention measurement point to be reached.

3 — Sample Size Calculation

Use the two-proportion z-test formula or a validated power calculator with:

  • Baseline rate: [actual value from data]
  • MDE: [minimum detectable effect, relative]
  • Power: 80% (or 90% for revenue decisions)
  • α: 0.05 (two-tailed)

Show the calculation. Do not use a default.

4 — Duration

= required sample per variant ÷ daily eligible traffic, rounded up to full weeks.

Apply rules:

  • Minimum 2 weeks for growth surfaces
  • Must exceed the retention measurement window for retention metrics
  • Maximum 8 weeks

If the required duration exceeds 8 weeks: ask whether to increase the MDE or find a higher-traffic surface.

5 — Stopping Rule

Fixed duration with optional conservative early-stop (p < 0.001 at ≥ 80% of target sample). No open-ended criteria.

Read the full file on GitHub · 78 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 78 lines · 58 tokens per session scan A 92b1a2661350

Subscribe to this mod's changes

experiment-designer is an agent published in the GitHub repository RBraga01/builder-growth (2 stars, last pushed 2mo ago), licensed MIT. It adds 58 tokens to every session and 743 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other agents, from other repositories

lumen

Owns the measurement layer — North Star definition, input metrics trees, A/B test specs, funnel and cohort analysis that drive product decisions. Use when defining what to track, diagnosing a funnel, or designing an experiment. Trigger with "define our North Star metric", "design an A/B test".

jeremylongshore/tons-of-skills-marketplace · 63 tokens

synthesis-worker

Reasoning-heavy synthesis worker for the Diffmode growth-tactics pipeline. Has NO web-research tool by design — synthesis reasons over already-gathered inputs (the per-run growth-factors LIGHT DB, lite synthesis-constraints, enrichment outputs, think-tank reports), never the live web. Loads a named synthesis-chain…

acogood/diffmode_free · 100 tokens

access-control

PostHog access control system implementation expert - use when adding access controls to new products, debugging access control issues, or questions about RBAC patterns.

PostHog/posthog · 32 tokens

report-writer

Turns finished research into client-ready deliverable specs. Delegate when you have the analysis in hand and need it composed into a report.yml or proposal.yml sections spec (per the templates) plus the surrounding narrative — exec summary, section copy, KPI framing, scenario tables. Precise and conservative: it…

prashishh/seo-geo-report-engine · 104 tokens

ds-agent-retention

Subagent that extracts subscription health and retention data from Stripe (and optionally internal database) via Dataslayer MCP. Returns structured findings only, no interpretation or recommendations.

Dataslayer-AI/Marketing-skills · 37 tokens

knowledge-synthesizer

Meta utility agent for multi-source intelligence synthesis. Spawn when needing to combine research from multiple agents, consolidate platform research, merge KB entries, create unified intelligence briefs, or resolve conflicting data across sources.

OpenAnalystInc/Vibe-Marketer · 45 tokens