investigate

Debugges software failures by collecting symptoms, testing hypotheses, and verifying fixes with regression proof.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you debug production or staging failures by enforcing root-cause investigation before any code change, preventing whack-a-mole fixes.

Core Features & Use Cases

  • Root-cause first (Iron Law): gathers symptoms and evidence, then formulates a specific, testable hypothesis before touching code.
  • 5-phase workflow with structured output: collects signals, locks scope (optional), tests hypotheses, implements the minimal fix, and verifies with regression proof.
  • 3-strike escalation: after three failed hypotheses, stops guessing and escalates for human review or adds instrumentation.

Use case: you deploy yesterday, and today a 500 error appears on a POST endpoint; run investigate to trace the regression, confirm the exact failing layer, and produce a DEBUG REPORT with a regression test.

Quick Start

Use /investigate with a clear description of the bug and its context (error message, endpoint, timeframe, and reproduction steps if known) to get a structured root-cause investigation plan and DEBUG REPORT.

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 500 error or stack trace in production?

To find the root cause of a 500 error, systematically collect symptoms, read relevant code paths, and test hypotheses before applying a minimal verified fix. This enforces strict root-cause validation and generates a structured debug report with regression test details.

What is the best way to debug intermittent production failures without guessing?

The best way to debug intermittent production failures is using a 3-hypothesis escalation mechanism. It tests three specific hypotheses against collected evidence, and if all fail, stops guessing and escalates for human review or adds further instrumentation.

How do I perform regression testing after fixing a failing software integration?

To perform regression testing after fixing a failing integration, implement a minimal verified fix and generate a structured debug report. This report includes the applied regression test details and evidence proving the root cause has been resolved.

Can I use hypothesis testing for unclear errors across web services and pipelines?

Yes, you can use hypothesis testing for unclear errors across web services and pipelines. The workflow gathers signals, locks scope, and tests specific hypotheses against the failing code paths to target regressions and trace the exact failing layer.

When should I escalate a failing software investigation instead of continuing to debug?

You should escalate a failing software investigation after three failed hypotheses. This 3-strike escalation rule stops the whack-a-mole cycle, halting further guessing and triggering a request for human review or additional instrumentation.