design-discriminating-prediction

SkillMedia

Lets your agent design claude skill experiments whose outcomes tell competing hypotheses apart.

Use design-discriminating-prediction in Claude, ChatGPT or Ahel Desktop

Free. Sign in, add design-discriminating-prediction and connect your AI. About a minute.

Claude ChatGPT Ahel Desktop

Also: Claude Code · Cursor · Codex

Then ask your AI: use the design-discriminating-prediction skill

Details

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Account requirements not reviewed. Check the skill instructions before use; Ahel provides instructions and does not run this skill.

No other account needed.

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

design-discriminating-predictionStart free

About this skill

Construct a prediction that yields different outcomes under competing hypotheses.

What this skill tells your AI

The instructions your AI receives, as published by yogsoth-ai/de-anthropocentric-research-engine in skills/design-discriminating-prediction/SKILL.md and read by Ahel’s review.

Purpose

Design an observation or test whose outcomes distinguish among competing hypotheses.

Input contract

required: [hypotheses, current_evidence, target_comparison]
optional: [constraints, measurable_variables, candidate_interventions]
constraints: [predictions must differ on an observable outcome]

Procedure

Identify the unresolved contrast between hypotheses.
Derive each hypothesis's prediction under candidate conditions.
Select the condition maximizing interpretable separation within constraints.
Specify outcome, measurement, and interpretation for each possible result.

Output contract

produces: [discriminating_test, prediction_table, measurement_plan, interpretation_rules]
delta_fields: [hypothesis_updates, decisions, uncertainties, recommended_jumps]

Quality gates

Predictions differ before the test is selected.
Confounds and ambiguous outcomes are recorded.

Failure and counterexamples

Do not call a test discriminating when all hypotheses predict the same result or when the measurement cannot separate them.

Provenance map

resolved: design-discriminating-prediction

Signals

GitHub stars: 503
Forks: 42
Last commit: Sep 2026

Advanced

Item type: skill
Key: design-discriminating-prediction
Source: github.com/yogsoth-ai/de-anthropocentric-research-engine

github.com/yogsoth-ai/de-anthropocentric-research-engine