hypothesis-check

hypothesis-check is a command for Claude Code from AInsteinsBR/renata. It costs 24 tokens per session (1,760 once invoked), scanned A, original, MIT.

A review command that compares each product requirement hypothesis with the real number measured after release. It gives every hypothesis its own clear verdict and next action.

In plain words
What is it for?
Use it after a measurable feature has run long enough to produce data, when a metric reaches its warning limit, or before starting the next major phase.
Why use it?
It prevents teams from treating an untested assumption as true or combining several results into one vague conclusion. It closes the loop between a planned target and what actually happened.

Command for Claude Code

Written for Claude Code: $ARGUMENTS substitution. Also seen: mentions CLAUDE.md.

Part of the renata plugin — 3 skills, 33 commands, 6 agents, 2 hooks shipped together

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.

agentmods
npx agentmods add commands/ainsteinsbr/renata/hypothesis-check
Clone the repo
git clone --depth 1 https://github.com/AInsteinsBR/renata

Made for: Claude Code.

Or install renata, the plugin that ships this one along with the rest of its 3 skills, 33 commands, 6 agents, 2 hooks.

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 hypothesis-check

README.md
[![agentmods](https://agentmods.dev/badge/commands/ainsteinsbr/renata/hypothesis-check.svg)](https://agentmods.dev/commands/ainsteinsbr/renata/hypothesis-check)
Your own site
<a href="https://agentmods.dev/commands/ainsteinsbr/renata/hypothesis-check"><img src="https://agentmods.dev/badge/commands/ainsteinsbr/renata/hypothesis-check.svg" alt="Measured on agentmods" height="20"></a>
Per session 24 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,760 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00024 $0.01760
Opus 5 $0.00012 $0.00880
Sonnet 5 $0.00005 $0.00352
Haiku 4.5 $0.00002 $0.00176

Measured 6d ago against content hash 5f1051341e4d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

hypothesis-check 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.

commands/hypothesis-check.md · 124 lines

How it starts

The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/renata:hypothesis-check — Confront the PRD hypotheses with real data (close the loop)

You are an honest, skeptical Product Manager. You take each PRD hypothesis (falsifiable, with baseline → target) and confront it with the real measured number. If the PRD has N hypotheses, each one gets its own verdict — never aggregate. You force an explicit verdict and, for each verdict, an action. It is the Measure-Learn step of the method — what closes the loop that /renata:prd + /renata:metrics opened.

Respond to the user and generate document content in the user's language (the language they are writing in).

Without this command, the hypothesis is born falsifiable and is never falsified. This command is the materialization of the principle "Evidence reopens decisions" (see METHOD.md › "The loop closes").

When to use

  • A phase delivered a measurable feature and has run long enough to have real data.
  • The kill criteria / tripwire of a metric (defined in /renata:metrics) was reached.
  • End of a phase with a product feature (/renata:retro points here).
  • Before starting the next big phase — confirm the previous bet held before doubling down on it.
  • You suspect you are building on top of a hypothesis that has already fallen.

Use /renata:hypothesis-check to falsify the bet (Measure-Learn). Use /renata:retro for the phase's execution learnings (what worked in the how). Use /renata:metrics to (re)define what to measure and the tripwire.

Before generating

  1. Read @docs/prd/ — extract each central hypothesis and its falsification signal. If the PRD has N intertwined hypotheses, each one gets its own verdict (do not aggregate).
  2. Read @docs/business-context/metricas.md — get the decisive metric, the baseline, the target, and the kill criteria.
  3. Read @docs/features/README.md — which features were delivered to move this metric (to evaluate sunset candidates).
  4. Read @CLAUDE.md (active phase) and the phase's doc.
  5. Hard precondition — ask and require:
    • What is the REAL measured number? (not estimated). If the user does not have the real data, STOP: the baseline/measurement is not instrumented. Do not invent a verdict. Instead, mark it with /renata:todo 🟡 "instrument metric X before the hypothesis-check" and guide them to come back when there is data.
    • Data source: where did the number come from? (analytics, query, survey). Without an auditable source, the verdict is faith.
    • Window: how much operating time does this number represent? (1 week of data does not close a 90-day retention hypothesis).

Read the full file on GitHub · 124 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 · 124 lines · 24 tokens per session scan A 5f1051341e4d

Subscribe to this mod's changes

hypothesis-check is a command published in the GitHub repository AInsteinsBR/renata (10 stars, last pushed 15d ago), licensed MIT. It adds 24 tokens to every session and 1,760 once invoked, about $0.0001 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.