d-verify

Load hypotheses in confidence order and validate them with debug logs.

Updated Jan 22, 2026

Search Tags:#debugging #investigation #logging #verification #root-cause #reproduction #hypotheses

Ready to run

npx skills add https://github.com/Hoang604/get-thing-done --skill d-verify

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: d-verify
Source: https://github.com/Hoang604/get-thing-done/tree/main/.claude/skills/d-verify
Command: npx skills add https://github.com/Hoang604/get-thing-done --skill d-verify

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill helps engineers systematically verify hypotheses to uncover the root cause of failures using structured reproduction steps and targeted debug logs.

Core Features & Use Cases

Load hypotheses in confidence order
Spawn verification tasks for each hypothesis
Generate a ROOT_CAUSE.md with evidence
Enforce evidence-based conclusions and stop if inconclusive

Quick Start

Place your reproduction steps in .gtd/debug/current/SYMPTOM.md and hypotheses in .gtd/debug/current/HYPOTHESES.md, then run the verification workflow.

Frequently Asked Questions about d-verify

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

FAQPage Schema

How do I verify debugging hypotheses with targeted logs to find a root cause?▼

To verify debugging hypotheses, load them in confidence order and validate each with strategic debug logs, generating a ROOT_CAUSE.md file with evidence once a hypothesis is confirmed.

What is the best way to structure reproduction steps for iterative debugging?▼

The best way to structure reproduction steps is to place them in a SYMPTOM.md file under .gtd/debug/current, alongside a HYPOTHESES.md file, enabling systematic verification of each potential root cause.

How does evidence-based root cause analysis work for software failures?▼

Evidence-based root cause analysis works by spawning verification tasks for each hypothesis, validating them with targeted debug logs, and enforcing evidence-based conclusions while stopping if results are inconclusive.

Do I need specific files to start systematic debugging and hypothesis testing?▼

Yes, you need to create SYMPTOM.md and HYPOTHESES.md files under the .gtd/debug/current directory, providing reproduction steps and ranked hypotheses to initiate the verification workflow.

What happens when debug log verification is inconclusive for a hypothesis?▼

When debug log verification is inconclusive, the process enforces evidence-based conclusions and stops, preventing unconfirmed hypotheses from being documented as the root cause in the output file.