decompose-global-sensitivity

SkillDev tools

Lets your agent estimate which uncertain inputs most affect a model's output using sensitivity analysis.

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 decompose-global-sensitivity skill

About this skill

Estimate global main effects and interactions across uncertain inputs using screening/decomposition methods such as Morris→Sobol or an equivalent supplied method.

What this skill tells your AI

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

Purpose

Estimate global main effects and interactions across uncertain inputs using screening and decomposition.

Input contract

required: [model, uncertain_inputs, output_metric]
optional: [input_distributions, screening_method, decomposition_method]
constraints: [input distributions and sampling design must be declared]

Procedure

  1. Define input ranges/distributions and the output metric.
  2. Screen influential variables with Morris or an equivalent declared method.
  3. Estimate main and interaction effects with Sobol or an equivalent variance decomposition.
  4. Compare sensitivity rankings across sampling assumptions.

Output contract

produces: [screening_report, main_effects, interaction_effects, sensitivity_ranking]
delta_fields: [findings, evidence_updates, uncertainties, decisions]

Quality gates

  • Main and interaction effects use a declared sampling design; rankings include uncertainty and do not conflate screening with decomposition.

Failure and counterexamples

Do not report global sensitivity when ranges are arbitrary, model evaluations are invalid, or interactions were omitted without notice.

Provenance map

  • deep-insight/variance-decomposition: resolved.
  • screening-then-decomposition: resolved.
  • morris-screening: resolved.
  • sobol-decomposition: resolved.

Signals

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