deep-dive

Trace ambiguous system behavior and produce evidence-backed requirements specifications.

1|Updated Apr 1, 2026
One-click install
npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill deep-dive-oliverouyang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-dive
Source: https://github.com/OliverOuyang/shuhe-work-skills/tree/main/skills/deep-dive
Command: npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill deep-dive-oliverouyang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Deep Dive helps teams uncover why a problem or unexpected behavior occurred and turns investigative findings into a clear, evidence-backed requirements specification so downstream planners and implementers can act without guessing.

Core Features & Use Cases

  • Two-stage pipeline: run a 3-lane causal trace followed by an interactive deep interview that crystallizes requirements.
  • 3-point trace→interview injection: injects trace synthesis, codebase context, and per-lane critical unknowns into the interview without treating data as executable instructions.
  • Artifact persistence & execution bridge: saves trace and spec artifacts, persists state for resume, and hands off spec files to omc-plan, autopilot, ralph, or team for execution.
  • Use Cases: bug root-cause investigations, ambiguous feature exploration, brownfield impact analysis, and preparing handoff-ready specs for autonomous execution.

Quick Start

Invoke the skill with a concise problem statement such as: /deep-dive "Production DAG fails intermittently on transformation step"

Frequently Asked Questions about deep-dive

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

FAQPage Schema
How do I investigate intermittent production bugs and generate a requirements specification?

To investigate intermittent production bugs and generate a requirements specification, provide a concise problem statement to trigger a 3-lane causal trace, followed by an interactive interview that crystallizes evidence-backed findings into a handoff-ready spec file.

What is the best way to analyze brownfield system impact and document requirements?

Analyzing brownfield system impact and documenting requirements is best handled by running parallel causal tracing to uncover root causes, then synthesizing codebase context and critical unknowns into an actionable specification for downstream planning.

Can I resume a root-cause analysis session if my investigation is interrupted?

You can resume an interrupted root-cause analysis session because the pipeline features artifact persistence and state saving, allowing you to pause and continue the causal trace and interactive interview without losing progress.

How do I turn ambiguous feature exploration findings into actionable specs?

To turn ambiguous feature exploration findings into actionable specs, the tool injects trace synthesis and per-lane critical unknowns into a stateful interview, directly translating investigative evidence into crystallized requirements.

Does the generated requirements specification work with autonomous execution tools?

The generated requirements specification works directly with autonomous execution tools by acting as an execution bridge, handing off the persisted spec files to downstream planners like omc-plan, autopilot, ralph, or team for immediate execution.