autoreskill-run-experiment

Launch, monitor, and reconcile AutoResearch experiments across multiple backends.

Updated May 27, 2026
One-click install
npx skills add https://github.com/Iranb/codex-autoresearch-skill-pack --skill autoreskill-run-experiment
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autoreskill-run-experiment
Source: https://github.com/Iranb/codex-autoresearch-skill-pack/tree/main/skills/autoreskill-run-experiment
Command: npx skills add https://github.com/Iranb/codex-autoresearch-skill-pack --skill autoreskill-run-experiment

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python, pandas, numpy, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill automates the launching and monitoring of experiments, eliminating manual intervention and ensuring reliable tracking and reconciliation of experiments.

Core Features & Use Cases

  • Experiment Launch and Monitoring: Automate the launch of experiments across various backends and monitor their progress.
  • Resource Management: Monitor resource budgets and prevent metric/dataset/baseline drift.
  • Reconciliation: Reconcile completed runs, update ledgers, and maintain best-known runs.
  • Use Case: When conducting experiments with complex parameter settings, this Skill ensures that the experiments are launched correctly and the results are tracked and recorded accurately.

Quick Start

Run the experiment with the desired parameters by executing the following command: autoreskill-run-experiment --project <project-root> --backend <backend> --command <command>

Frequently Asked Questions about autoreskill-run-experiment

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

FAQPage Schema
How do I automate experiment launches across remote HPC backends?

Automate experiment launches across remote HPC backends by configuring the tool to connect via SSH, AutoDL, or BJTU HPC. It handles execution, monitors progress, records logs, and updates ledgers automatically.

What is the best way to prevent metric drift during automated experiment tracking?

Preventing metric drift during automated experiment tracking is achieved through resource budget monitoring and reconciliation. The tool updates ledgers and maintains best-known runs to ensure dataset and baseline consistency.

How do I launch an AutoResearch experiment from the command line?

To launch an AutoResearch experiment from the command line, execute `autoreskill-run-experiment --project <project-root> --backend <backend> --command <command>` to start and monitor the run automatically.

Do I need Python and Pandas to run experiment monitoring scripts?

Yes, you need Python, Pandas, and NumPy installed to run experiment monitoring scripts. These dependencies are required for executing commands, tracking metrics, and reconciling experiment ledgers.

Does this tool support local experiment execution or only remote backends?

This tool supports both local experiment execution and remote backends. You can automate launching, monitoring, and reconciliation across local environments, SSH, AutoDL, and BJTU HPC setups.

Why does experiment reconciliation fail when updating ledgers?

Experiment reconciliation fails when updating ledgers if resource budgets are exceeded or metric drift is detected. The tool prevents these failures by monitoring progress and maintaining best-known runs.