Skill

validate-hypothesis

Validates pending design hypotheses for features by invoking hypothesis-tester agent with code experiments, codebase analysis, external research; handles rejections via user overrides.

Bash

design

testing

npx claudepluginhub rp1-run/rp1 --plugin rp1-dev

Tool Access

This skill is limited to using the following tools:

Bash(echo *)Bash(rp1 *)

Preview

Invokes **hypothesis-tester** agent to validate design assumptions.

SKILL.md

Similar Skills

validate-hypothesis

Validates pending design hypotheses in features/hypotheses.md via rp1-dev:hypothesis-tester agent using code experiments, codebase analysis, and research. Handles rejections with user overrides.

2 tools

rp1-dev

define-hypothesis

178

Defines testable hypotheses with success metrics, target users, and validation approaches for experiment design, A/B tests, and team alignment on assumptions.

2 files

pm-skills

hf-experiment

Generates structured probe plans to validate blocking, high-risk, or low-confidence key assumptions in product discovery and spec phases before advancing to implementation.

1 file

harness-flow

Stats

Parent Repo Stars13

Parent Repo Forks3

Last CommitMar 17, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Hypothesis Validator

Invokes hypothesis-tester agent to validate design assumptions.

Parameters

Extract these parameters from the user's input:

Parameter	Required	Default	Description
`FEATURE_ID`	Yes	-	Feature identifier whose hypotheses to validate (kebab-case)

Environment values (resolve via shell):

RP1_ROOT: !rp1 agent-tools rp1-root-dir (extract data.root from JSON response)

Prerequisites

{{$RP1_ROOT}}/work/features/{FEATURE_ID}/hypotheses.md MUST exist
Generated by feature-design when high-risk assumptions identified

Execution

Step 1: Invoke Agent

Task tool: subagent_type: rp1-dev:hypothesis-tester prompt: FEATURE_ID: {FEATURE_ID} RP1_ROOT: {{$RP1_ROOT}} Validate all PENDING hypotheses for this feature.

Agent actions: load hypotheses.md -> parse PENDING -> validate via experiment/analysis/research -> document findings w/ evidence -> update status CONFIRMED|REJECTED -> cleanup temp artifacts -> report summary

Step 2: Handle Rejected Hypotheses

Parse agent output. If JSON block w/ type: "rejected_hypotheses":

{
  "type": "rejected_hypotheses",
  "hypotheses": [...],
  "hypotheses_path": "..."
}

For each rejected:

AskUserQuestion: "{id} REJECTED: {statement}. Evidence: {evidence_summary}. Domain knowledge confirms valid?" Options:

Accept rejection
Override - I confirm valid

If "Override":

Edit hypotheses.md:
- Status: REJECTED -> CONFIRMED_BY_USER
- Append: **User Override**: User confirmed validity based on domain knowledge.

If "Accept rejection": No change (status remains REJECTED)

Step 3: Report Result

Display validation summary. Note any user overrides.