AI Debugging Tool: Evidence-Based ML Experiment Diagnosis

AI Debugging Tool: Evidence-Based ML Experiment Diagnosis

The AI Debugging Tool: Evidence-Based ML Experiment Diagnosis is designed to assist in diagnosing failing machine learning experiments by employing a systematic approach. This skill probes the system to collect empirical evidence before forming hypotheses or taking any action, reducing the likelihood of improper troubleshooting steps that could obscure the real issues.

What this skill does

• Conducts an initial set of cost-effective probes to gather immediate data, such as process state, kernel and system events, GPU statistics, and log scrolling to evaluate the current state of an ML process.
• Forms hypotheses based on the gathered evidence rather than plausible guessing, ensuring measured and calculated troubleshooting strategies.
• Executes smoke tests to validate these hypotheses before determining a root cause, allowing for cautious progression through the debugging process.
• Applies controls to manage the environment and only then claims a diagnosed issue and cause, based on sequential and methodological investigation.
• Integrates with AI coding agents like Claude Code, Cursor, and Codex.

Who it is for

This skill is invaluable for developers and technical teams utilizing AI coding agents who require precise and evidence-based diagnostic procedures. It supports individuals responsible for troubleshooting complex ML training issues and is suited for scenarios where conventional debugging methods prove inadequate.

Use cases

• When an ML process is identified as failing, diverging, running out of memory (OOM), or exhibiting unexpected behavior.
• When metrics appear abnormal, such as loss ratios increasing without cause or GPU utilization registering as zero, prompting an in-depth review.
• For situations requiring evidence-based analysis upon receiving log snippets with queries about operational failures and bugs.

Technical details

This AI Debugging Tool relies on a structured probing method that collects real-time data through system-level commands and checks, such as process monitoring using Linux commands like ps aux and dmesg. The disciplined sequence of probe, hypothesis, smoke test, control, and conclusion ensures that diagnostic conclusions are made carefully and supported by evidence.

Source & Licence

This package is built on open-source work published by fcakyon (fcakyon/phd-skills) and distributed under MIT. The original licence text and copyright notice are included in your download.

Personal and commercial use, modification and redistribution are permitted, provided the original copyright and licence notice are retained.

Your purchase covers curation, licence verification, packaging, documentation and instant delivery. It does not grant exclusive rights to the underlying open-source code, which remains available under its original licence.

Delivery & Support

• Delivery: instant — a secure download link is emailed to you as soon as payment is confirmed.
• Format: ZIP archive containing the skill files, documentation and the original licence.
• Support: support@mcpcart.com — we aim to reply within 2 business days.
• Updates: updates are included only where stated on this page.

Refunds

This is a digital product delivered immediately after purchase. By completing your order you request immediate delivery and acknowledge that, once the download has been accessed, the statutory right to cancel no longer applies to the extent permitted by law. Refund requests are handled in accordance with our published Refund Policy.

Claude, Codex, Gemini and Cursor are trademarks of their respective owners. MCP Cart is an independent marketplace and is not affiliated with, endorsed by, or sponsored by any of them. Compatibility references describe interoperability only.