Monitor Experiment

wanshuiyin/auto-claude-code-research-in-sleep

This pulls results from running ML experiments across SSH servers, Vast.ai instances, or Modal apps. It scrapes screen sessions, fetches JSON outputs, pulls W&B metrics when configured, and presents everything in comparison tables with deltas against baseline. The external cadence pattern is front and center here: it's designed to loop on a timer, check machine-verifiable completion signals like exit codes or file presence, then bail out without judging quality. Includes cost awareness reminders for Vast.ai billing and Feishu notifications if you've got that wired up. Useful when you've kicked off a grid search overnight and want structured status without SSH-ing in manually.

skills.sh installs: 547
GitHub stars: 18.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add wanshuiyin/auto-claude-code-research-in-sleep --skill monitor-experiment --agent claude-code

Installs into .claude/skills of the current project.

Files

Monitor Experiment Results

⏱ External cadence is appropriate here. This skill waits on an external fact (job completion / progress), so it is a natural /loop / CronCreate surface: the wake reads status and self-judges only machine-checkable completion (exit code, file exists, epoch logged) — never quality. This is the additive external-wait shape in shared-references/external-cadence.md. If a scheduled wait here ends in a verdict step (e.g. then audit results), run that verdict once after the wait clears — not re-entered per tick.

Monitor: $ARGUMENTS