Evaluate experiment results against research claims and route next steps.
npx skills add https://github.com/jkfee/Auto-Research --skill result-to-claim-jkfee
Please help me install this Agent Skill. Skill: result-to-claim Source: https://github.com/jkfee/Auto-Research/tree/main/skills/result-to-claim Command: npx skills add https://github.com/jkfee/Auto-Research --skill result-to-claim-jkfee
Experiments produce numbers; this gate decides what those numbers mean and helps determine whether results support the intended claims, identify gaps, and decide next actions before writing papers or running ablations.
Feed your completed experiment results into Codex for evaluation and routing.
High-intent search queries and answers about installing and using this skill.
To evaluate if experiment results support research claims, this skill collects data from wandb, EXPERIMENT_LOG.md, and log files to generate a structured verdict. It identifies supported claims, missing evidence, and suggests claim revisions.
The best way to route next steps after analyzing experiment data is by using a normalized Codex verdict. It routes your project to pivot, supplement, or confirm actions based on whether results provide no, partial, or full support.
Yes, you can use wandb run history to check experiment integrity before writing papers. The skill parses wandb data and EXPERIMENT_AUDIT.json to provide a confidence score and integrity status for your research contract.
To identify missing evidence and suggest claim revisions from research contracts, the skill compares results against docs/research_contract.md. It outputs specific fields detailing what results do not support and what next experiments are needed.
Yes, this workflow records reasoning for why experiment results do not support claims. It documents the detailed evaluation logic and routing decisions directly into findings.md for future reference.
jandan138Collect experiment results and route claim judgments via Codex MCP.
LingrongyeEvaluate experimental results against claims and route to pivot, supplement, or confirm.
x-TheFoxLog ML experiments, sweeps, and artifacts with WandB.
Mang30Generate a claim-driven experiment roadmap with run order and compute estimates.
