speckit.product-forge.experiment-design

speckit.product-forge.experiment-design is a command for Claude Code from VaiYav/speckit-product-forge. It costs 132 tokens per session (2,149 once invoked), scanned A, original, MIT.

A planning command for controlled A/B experiments, where different users see different versions of a feature. It defines the expected outcome, measurements, sample size, exposure rules, and decision thresholds before the feature is released.

In plain words
What is it for?
Use it to prepare a pre-registered experiment for a feature flag, including primary, secondary, and safety measurements.
Why use it?
It reduces the risk of changing the success criteria after seeing the results. When configured with a connected PostHog service, it can also create the corresponding feature flag and experiment.

Command for Claude Code

Written for Claude Code: $ARGUMENTS substitution.

Part of the speckit-product-forge plugin — 31 commands shipped together

Good fit Use it to prepare a pre-registered experiment for a feature flag, including primary, secondary, and safety measurements.

Compare 6 commands from other repositories ↓
Install with agentmods
npx agentmods add commands/vaiyav/speckit-product-forge/experiment-design
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/VaiYav/speckit-product-forge

Made for: Claude Code.

Or install speckit-product-forge, the plugin that ships this one along with the rest of its 31 commands.

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 speckit.product-forge.experiment-design

README.md
[![agentmods](https://agentmods.dev/badge/commands/vaiyav/speckit-product-forge/experiment-design/github.svg)](https://agentmods.dev/commands/vaiyav/speckit-product-forge/experiment-design)
Your own site
<a href="https://agentmods.dev/commands/vaiyav/speckit-product-forge/experiment-design"><img src="https://agentmods.dev/badge/commands/vaiyav/speckit-product-forge/experiment-design/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for speckit.product-forge.experiment-design

Your own site · 80×15
<a href="https://agentmods.dev/commands/vaiyav/speckit-product-forge/experiment-design"><img src="https://agentmods.dev/badge/commands/vaiyav/speckit-product-forge/experiment-design.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 132 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,149 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 $0.00132 $0.02149
Opus 5 $0.00066 $0.01074
Sonnet 5 $0.00026 $0.00430
Haiku 4.5 $0.00013 $0.00215

Measured 9d ago against content hash 5b012e9e3ee9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

speckit.product-forge.experiment-design 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 9d 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.

commands/experiment-design.md · 281 lines

How it starts

The opening of the file, as written. The whole thing — 281 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Product Forge — Experiment Design (Phase 9B)

You are the Experimentation Analyst for Product Forge. Your job: force an honest, pre-registered A/B plan before the feature ships, so results cannot be re-interpreted to fit a desired outcome.

This phase is opt-in and applies only when a feature ships behind a flag with experimentation enabled.

User Input

$ARGUMENTS

Parse for:

  • Feature slug (required).
  • --flag=<key> — the feature-flag key used for exposure (cross-checked against flags/registry.yml from release-readiness).
  • --variants=<list> — comma-separated variant names. Default: control,treatment.

Real experiment (v1.6, Theme D): when telemetry.product_analytics is posthog and its MCP is connected, optionally create the actual flag + experiment (with the pre-registered primary/guardrail metrics) via the PostHog MCP after the user pre-registers the plan. The retrospective then reads this experiment's real results. Without a connected MCP, produce the plan artifacts only.


Step 0: Prerequisites

  1. .forge-status.yml has release_readiness.status completed.
  2. flags/registry.yml exists and contains the specified flag.
  3. research/metrics-roi.md exists (source of primary metric expectations). If missing, ask the user for a primary metric; do not invent one.
  4. Analytics provider has been decided (from project config or tracking-plan).

Step 1: Hypothesis Statement

Produce a one-sentence hypothesis in this shape:

Because {user insight from research}, if {we ship the treatment variant}, then {primary metric} will move by at least {MDE} in the direction {up / down} within {measurement window}.

Example:

Because new users abandon before activation when the onboarding asks for too much up front, if we defer the avatar selection to after first chat, then Day-1 activation rate will move up by at least 3 percentage points within 14 days.

Reject vague hypotheses ("X will improve engagement"). If the user cannot commit to a direction and a number, stop and flag it as a gap.

Read the full file on GitHub · 281 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. 9d ago First seen · 281 lines · 132 tokens per session scan A 5b012e9e3ee9

Subscribe to this mod's changes

speckit.product-forge.experiment-design is a command published in the GitHub repository VaiYav/speckit-product-forge (23 stars, last pushed 16d ago), licensed MIT. It adds 132 tokens to every session and 2,149 once invoked, about $0.0007 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 2026-08-30.