skills.rest
OpenLAIR
OpenLAIROFFICIAL
Explore all skills by this author →
GitHub

aris-result-to-claim

Evaluates experiment results against intended claims and routes to next research action.

1.1k|116|Updated Feb 26, 2026
Search Tags:#WandB#claim validation#research workflow#codex mcp#experiment evaluation#ablation planning#ML experiments
One-click install
npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-result-to-claim
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aris-result-to-claim
Source: https://github.com/OpenLAIR/dr-claw/tree/main/skills/aris-result-to-claim
Command: npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-result-to-claim

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

After experiments finish, researchers often over-claim what their results support. This Skill objectively judges whether experimental results actually back the intended claim, identifies missing evidence, and routes the project to the right next step (pivot, supplement, or confirm) before paper writing begins.

Core Features & Use Cases

  • Result Collection: Gathers experiment data from W&B runs, EXPERIMENT_LOG.md, EXPERIMENT_TRACKER.md, training logs, and research contracts.
  • Codex-Based Judgment: Sends results to Codex MCP for an objective verdict (yes/partial/no) on claim support, with confidence levels and suggested claim revisions.
  • Automatic Routing: Based on the verdict, records postmortems, updates claims, triggers ablation planning, or recommends pivoting to new ideas; optionally updates a research wiki with experiment-claim edges.
  • Use Case: After training runs complete on three datasets, invoke this Skill to check whether the results support the paper's central claim before drafting the experiments section.

Quick Start

Run the result-to-claim gate on my latest W&B run to judge whether the results support my intended claim.

Frequently Asked Questions about aris-result-to-claim

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

FAQPage Schema
How do I check if my experiment results support my paper claim?▼

Collect results from W&B, experiment logs, or training logs, then submit them with the intended claim to Codex for a structured verdict. The judgment returns claim_supported (yes/partial/no), missing evidence, and suggested claim revisions.

How to evaluate ML experiment results with W&B data?▼

Fetch metrics and training curves via wandb.Api().run(...).history(), assemble baseline comparisons and deltas, then send the summary for objective evaluation. The workflow also reads EXPERIMENT_LOG.md and EXPERIMENT_TRACKER.md when W&B is unavailable.

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

A partial verdict updates the working claim to reflect what is supported, records the gap in findings.md, and triggers supplementary experiments. Multiple partial rounds on the same claim prompt narrowing the claim scope or switching ideas.

Can this workflow run if Codex MCP is unavailable?▼

Yes. If the Codex MCP call fails, the assistant makes its own judgment and marks it as pending Codex review, so the research pipeline is not blocked. The verdict and reasoning are still recorded in findings.md.

When should I run a result-to-claim evaluation in my research pipeline?▼

Run it after main experiments complete but before writing the paper or running ablations. It is designed for ambiguous results needing an objective second opinion, not for sanity checks.

Related Skills

Explore More Education & Research Skills→
duypham2801duypham2801

result-to-claim

Evaluate experimental results against research claims using Codex verdicts.

Community
Advanced
ShiienShiien
7

writing-outline

Draft research paper outlines with section structure and a Claims-Evidence Matrix.

Community
Intermediate
LambenthanLambenthan
77

exp-eval

Judge completed experiments and update wiki claims with verdicts.

Community
Advanced
nguyenphpnguyenphp
41

ab-test-dashboard

Create and analyze A/B test dashboards with statistical significance calculations.

Community
Intermediate