skills.rest
Lingrongye
LingrongyeCOMMUNITY
Explore all skills by this author →
GitHub

result-to-claim

Evaluate experimental results against claims and route to pivot, supplement, or confirm.

1|Updated Mar 26, 2026
Search Tags:#routing#codex#evidence#claims#experiments#evaluate#judgment
One-click install
npx skills add https://github.com/Lingrongye/federated-learning --skill result-to-claim-lingrongye
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: result-to-claim
Source: https://github.com/Lingrongye/federated-learning/tree/main/Auto-claude-code-research-in-sleep/skills/result-to-claim
Command: npx skills add https://github.com/Lingrongye/federated-learning --skill result-to-claim-lingrongye

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Experiments generate numbers and claims, but teams need a clear verdict on what the results actually support. This Skill provides a structured gate that evaluates results against intended claims and directs next steps (pivot, supplement, or confirm) using Codex judgments.

Core Features & Use Cases

  • Collect Results: Gather metrics and evidence from available sources such as wandb histories, EXPERIMENT_LOG.md, EXPERIMENT_TRACKER.md, training logs, and docs/research_contract.md to capture the full experimental context.
  • Codex Judgment: Send the assembled information to Codex for an objective evaluation and produce a verdict with guidance on next actions.
  • Parse and Normalize: Extract structured fields (e.g., claim_supported, what_results_support, what_results_dont_support, missing_evidence, suggested_claim_revision, next_experiments_needed, confidence) for consistent routing.
  • Route Based on Verdict: Apply the verdict to determine post-processing steps: no -> postmortem and pivot; partial -> refine claims and run supplementary experiments; yes -> mark claim as supported and proceed to paper drafting.

Quick Start

After experiments finish, run the result-to-claim workflow to obtain Codex judgment and auto-route the results.

Frequently Asked Questions about result-to-claim

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

FAQPage Schema
How do I evaluate experimental results against intended claims?▼

Evaluating experimental results against intended claims involves collecting metrics from sources like wandb histories and EXPERIMENT_LOG.md, then applying Codex judgment to produce a structured verdict on whether your data supports the claims.

What is the best way to route next actions after an experiment completes?▼

Routing next actions after an experiment completes requires parsing a Codex judgment verdict into structured fields like claim_supported and confidence, which automatically directs your workflow to pivot, run supplementary experiments, or proceed to paper drafting.

Can I use wandb histories and research contracts to determine if results support claims?▼

Yes, you can use wandb histories and docs/research_contract.md to determine if results support claims by assembling the experimental context and sending it to Codex for an objective evaluation that outputs structured evidence and next steps.

How do I parse and normalize experimental evidence for consistent routing?▼

To parse and normalize experimental evidence for consistent routing, extract structured fields such as what_results_support, missing_evidence, and suggested_claim_revision from the Codex evaluation to standardize the post-processing workflow.

What happens when experimental results only partially support a claim?▼

When experimental results only partially support a claim, the evaluation verdict routes the workflow to refine the claims and run supplementary experiments to gather the missing evidence needed for confirmation.

Related Skills

Explore More Data & Analytics Skills→
jandan138jandan138

result-to-claim

Collect experiment results and route claim judgments via Codex MCP.

Community
Advanced
HeXiao-55HeXiao-55
1

result-to-claim

Evaluate experimental results against claims and route next steps.

Community
Advanced
cabeznocabezno

weights-and-biases

Log ML experiment metrics and manage runs, artifacts, and sweeps in W&B.

Community
Intermediate
XWIlluDeluXWIlluDelu

scientific-critical-thinking

Evaluate scientific claims by critiquing methodology, bias, and statistical validity.

Community
Advanced