skills.rest
ZK-Theory
ZK-TheoryOFFICIAL
Explore all skills by this author →
GitHub

sensitivity-comparison-review

Verifies sensitivity-analysis and comparison result JSONs retain all fields required by downstream comparison tables.

1|Updated Dec 13, 2025
Search Tags:#sensitivity analysis#json validation#schema contract#comparison table#field presence#result review#statistical output
One-click install
npx skills add https://github.com/ZK-Theory/TDL --skill sensitivity-comparison-review-zk-theory
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sensitivity-comparison-review
Source: https://github.com/ZK-Theory/TDL/tree/main/.agents/skills/sensitivity-comparison-review
Command: npx skills add https://github.com/ZK-Theory/TDL --skill sensitivity-comparison-review-zk-theory

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Sensitivity-analysis and comparison result JSONs can validate as well-formed while silently dropping fields that downstream comparison tables and figures need, forcing costly re-runs of the analysis to recover missing data. ## Core Features & Use Cases - Downstream-field inventory: Lists every field the comparison tables and figures consume, such as test statistic T, dimension d, mean, p-value, and per-cell identifiers. - Field presence and comparability checks: Confirms each required field is present for every arm, cell, or probe, on the same scale and computed the same way across arms. - Silent narrowing detection: Catches schema changes that drop fields, preferring a schema contract with explicit required_keys over informal review. - Use Case: When reviewing LM-sensitivity result JSONs before building a comparison table, run this review to get a field-by-field PRESENT / MISSING report across all arms and confirm the table can be built from the JSON as-is. ## Quick Start Review my sensitivity-analysis result JSONs and confirm every field needed for the comparison table is present across all arms.

Frequently Asked Questions about sensitivity-comparison-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify comparison result JSONs have all required fields?▼

Inventory every field the downstream comparison tables and figures consume, then confirm each is present in the result JSON for every arm, cell, or probe. Report field-by-field PRESENT or MISSING status across arms.

How to prevent schema changes from silently dropping result fields?▼

Define a schema contract with explicit required_keys for the result JSON instead of relying on informal review. This catches field drops at validation time rather than when the comparison table fails to build.

Why does a well-formed JSON still break my comparison table?▼

A JSON can be syntactically valid while missing fields the table consumes, such as test statistic T, dimension d, or mean. Structural validity does not guarantee the presence of downstream-required content.

What makes fields comparable across analysis arms?▼

Fields must be on the same scale and computed the same way across arms, with per-arm identifiers that let rows be matched. Without this, present fields still cannot be assembled into a valid comparison table.

When should I use a schema contract instead of manual review?▼

Use a schema contract whenever result JSONs feed automated downstream consumers like comparison tables. Manual review catches issues once; a contract with required_keys mechanizes the requirement and catches regressions on every run.

Related Skills

Explore More Data & Analytics Skills→
JYeswakJYeswak

stat-review

Reviews experiment analyses for power, multiplicity, and assumption violations before shipping decisions.

Community
Intermediate
LightconeResearchLightconeResearch
18

figure-comparison

Compares paper reference figures, tables, and values with reproduced results.

Official
Advanced
synaptiaisynaptiai
4

delegate

Delegate tasks to subagents with explicit contracts and merge strategies.

Official
Advanced
flutterflutter
2.9k

api-review

Reviews code against canonical API design guidelines and generates a graded report.

Official
Intermediate