classify-falsification-verdict

SkillDev tools

Lets your agent classify a tested claim as BROKEN, CORROBORATED, or UNFALSIFIABLE based on falsification test results.

Available today. Use it from your connected AI after setup.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the classify-falsification-verdict skill

About this skill

Classify a tested claim as BROKEN, CORROBORATED, or UNFALSIFIABLE based on whether a legitimate falsifier succeeded, failed under adequate power, or cannot be specified/reached.

What this skill tells your AI

The instructions your AI receives, as published by yogsoth-ai/de-anthropocentric-research-engine in skills/classify-falsification-verdict/SKILL.md and read by ahel’s review.

Purpose

Classify a tested claim as BROKEN, CORROBORATED, or UNFALSIFIABLE.

Input contract

required: [claim, falsifier, test_result, adequacy_assessment]
optional: [scope, power_or_precision]
constraints: [falsifier must be legitimate and within scope]

Procedure

  1. Check the falsifier and scope.
  2. Assess whether the test was adequate.
  3. Return exactly one permitted verdict with rationale.

Output contract

produces: [falsification_verdict, verdict_rationale, adequacy_assessment]
delta_fields: [findings, evidence_updates, uncertainties, decisions]

Quality gates

  • BROKEN requires a successful legitimate falsifier.
  • CORROBORATED requires an adequate test that failed to falsify.
  • Otherwise return UNFALSIFIABLE.

Failure and counterexamples

Do not treat an underpowered or unreachable test as corroboration.

Provenance map

  • resolved: falsification-first-stress-test

Signals

GitHub stars
501
Forks
41
Last commit
Sep 2026
Advanced
Catalog kind
skill
Key
classify-falsification-verdict
Source
github.com/yogsoth-ai/de-anthropocentric-research-engine