hunt

Diagnose root causes of errors and failing tests with evidence-based hypotheses.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/treejh/ai-agent-skills --skill hunt-treejh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hunt
Source: https://github.com/treejh/ai-agent-skills/tree/main/skills/hunt
Command: npx skills add https://github.com/treejh/ai-agent-skills --skill hunt-treejh

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill enforces diagnosing the true root cause of errors, crashes, unexpected behavior, and failing tests before any code changes are made, preventing symptom-only patches and regression cascades.

Core Features & Use Cases

  • Evidence-first diagnosis: Require a one-sentence, testable hypothesis that names a specific file, function, line, or condition and cites supporting evidence.
  • Progress signals & instrumentation: Recommend a single targeted instrument (log line, failing assertion, or minimal test) to confirm or refute the hypothesis before applying a fix.
  • Hard safety rules: Enforce environment verification, grep-before-edit, stop after three failed hypotheses with user escalation, and avoid editing from memory.
  • Use Case: When a CI test suddenly fails or a runtime crash appears, run this Skill to trace execution paths, validate hypotheses with concrete evidence, and produce a minimal, testable fix plan.

Quick Start

Ask the hunt skill to diagnose the root cause of the failing test and propose one minimal instrument to confirm it before making any code changes.

Frequently Asked Questions about hunt

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

FAQPage Schema
How do I find the root cause of a failing test before applying a code fix?

To find the root cause of a failing test, formulate a one-sentence testable hypothesis citing specific files or lines, then use targeted instrumentation like log lines or minimal tests to gather evidence before making code changes.

What is the best way to debug unexpected crashes during incident triage?

The best way to debug unexpected crashes during incident triage is to validate hypotheses with concrete evidence, enforce environment verification, and avoid random code changes to prevent symptom-only patches and regression cascades.

How do I stop making random code changes when diagnosing errors?

To stop making random code changes when diagnosing errors, enforce a three-failed-hypotheses stop limit with user escalation, and require grep-before-edit checks to validate assumptions against the actual codebase.

Can I use an evidence-based approach for continuous integration failures?

Yes, you can use an evidence-based approach for continuous integration failures by tracing execution paths, proposing a single targeted instrument to confirm hypotheses, and producing a minimal, testable fix plan.

Why should I verify my environment before debugging runtime crashes?

You should verify your environment before debugging runtime crashes to ensure the issue is reproducible and isolated, preventing wasted effort on environment-specific discrepancies rather than true code defects.

Does evidence-based debugging work for local development and production incidents?

Evidence-based debugging works for local development and production incidents by applying consistent safety rules like grep-before-edit and hypothesis validation to trace execution paths and gather concrete proof.