flow-investigate

Diagnose software issue root causes through iterative user-guided hypothesis experiments.

3|Updated Oct 5, 2025
One-click install
npx skills add https://github.com/korchasa/flow --skill flow-investigate
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: flow-investigate
Source: https://github.com/korchasa/flow/tree/main/framework/skills/flow-investigate
Command: npx skills add https://github.com/korchasa/flow --skill flow-investigate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Diagnoses root causes through an iterative, user-guided hypothesis and experiment workflow, enabling safer debugging.

Core Features & Use Cases

  • Hypothesis generation with probabilistic ranking
  • Experiment design and approval gating
  • Hypothesis board visualization and iteration tracking
  • Safe baseline restoration between experiments
  • Applicable to complex debugging in software projects

Quick Start

Initiate an investigation session and begin by listing three likely hypotheses.

Frequently Asked Questions about flow-investigate

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

FAQPage Schema
What is iterative root-cause investigation for software debugging?

Root-cause investigation diagnoses software issues by generating hypotheses, designing experiments, and gating decisions on user approval to safely identify underlying causes.

How do I track debugging hypotheses and experiments during root-cause analysis?

You can track debugging iterations using a task-management tool that presents a hypothesis board, ranks hypotheses, and logs experiment decisions to maintain a structured root-cause analysis workflow.

Can I restore a safe baseline between debugging experiments?

Yes, the workflow supports safe baseline restoration between experiments, ensuring your software environment returns to a known state before testing the next ranked debugging hypothesis.

What's the best way to start a root-cause investigation for a complex software issue?

The best way to start a root-cause investigation is to initiate a session and list three likely hypotheses, which are then probabilistically ranked to guide subsequent experiment design and user approval.

When should I use a hypothesis-driven workflow for debugging?

You should use a hypothesis-driven workflow for debugging complex software projects where unstructured troubleshooting is insufficient and experiments require controlled iteration, user-gated decisions, and baseline restoration.