investigate

Execute diagnostic actions to confirm root causes with observed evidence.

Updated Feb 23, 2026
One-click install
npx skills add https://github.com/eumemic/dev-skills --skill investigate-eumemic
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: investigate
Source: https://github.com/eumemic/dev-skills/tree/main/skills/investigate
Command: npx skills add https://github.com/eumemic/dev-skills --skill investigate-eumemic

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps debug and diagnose broken behavior by driving a systematic, evidence-first investigation loop until the root cause is confirmed rather than guessed.

Core Features & Use Cases

  • Investigation loop: Continuously pick the highest-ROI action, execute it with targeted queries/data examination, and return to refine hypotheses with evidence.
  • Evidence over theorizing: Prioritizes confirming facts from actual data (code paths, logs, reproducible behavior) before proposing fixes or writing tests.
  • Mystery tracking: Converts newly discovered inconsistencies into explicit follow-up mysteries so nothing important is lost when context shifts.
  • Use case: A recurring failure after multiple “fix attempts” where you need to determine why the system behaves that way and what change will genuinely resolve it.

Quick Start

Use the investigate skill to diagnose the failing behavior and identify the confirmed root cause with a stepwise evidence loop.

Frequently Asked Questions about investigate

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

FAQPage Schema
How do I find the root cause of a bug when previous fix attempts failed?

Yes, evidence-based debugging handles multiple unknowns by converting newly discovered inconsistencies into explicit follow-up mysteries. This mystery tracking preserves context so unresolved questions are systematically investigated rather than lost during context shifts.

What is the best way to debug an issue using an evidence-first investigation loop?

Root cause analysis prioritizes drilling into actual content rather than metadata. You must confirm hypotheses with observed data from actual code paths, logs, and reproducible behavior, iterating until the root cause is fully evidenced or constraints block further progress.

Can I use this systematic debugging approach for recurring system failures with multiple unknowns?

Yes, evidence-based debugging handles multiple unknowns by converting newly discovered inconsistencies into explicit follow-up mysteries. This mystery tracking preserves context so unresolved questions are systematically investigated rather than lost during context shifts.

Why does guessing the fix fail to resolve unexpected behavior in complex systems?

Guessing fails because it skips confirming facts from actual data. Evidence-first investigation prioritizes examining actual code paths, logs, and reproducible behavior to guarantee the proposed change genuinely resolves the unexpected behavior.