How long does a Deep Interview session take?

Sessions typically last between 8 to 15 rounds of questioning, depending on the complexity of the feature and the clarity of the initial description.

What is the output of the Deep Interview skill?

The primary output is a detailed @spec.md file that contains the goals, data models, UI flows, and edge cases needed to guide implementation agents.

When should I use the Deep Interview skill?

Use it for complex features touching multiple screens, new data models, or any architectural decision where a 'just build it' approach might lead to significant rework later.

Does this skill only work for iOS development?

While it includes specific taxonomies for iOS (like Swift state management), the core pattern of requirements gathering and spec generation is valuable for any software project.

What is a 'Deep Interview' in the context of Claude Code?

A Deep Interview is a structured dialogue where the AI asks a series of targeted questions to fully map out requirements, edge cases, and architecture before any code is written.

Deep Interview & Spec Architect

Name: Deep Interview & Spec Architect
Author: willsigmon

bywillsigmon

•

Mobile Development

Conducts rigorous requirements interviews to generate comprehensive technical specifications and prevent project rework.

Deep Interview is a spec-first development skill that guides users through a structured discovery process to uncover hidden requirements before writing code. By 'grilling' the user with 8-15 rounds of pointed, domain-specific questions—particularly optimized for iOS development—it identifies core invariants, data relationships, and edge cases. The process culminates in a polished @spec.md file, ensuring that the subsequent implementation phase is grounded in a single, well-vetted source of truth, thereby eliminating expensive architectural rework.

Key Features

01iOS-Specific Taxonomy: Specialized focus on @Observable state, CloudKit sync, and mobile edge cases.

02Intelligent Question Clustering: Groups 1-3 related questions per round to maintain context without overwhelming the user.

03Automated Spec Generation: Produces a structured 10-section @spec.md including data models and error handling.

046 GitHub stars

05Integration-Ready: Seamlessly hands off approved specifications to implementation agents for execution.

06Iterative Questioning: Conducts 8-15 rounds of deep inquiry to stress-test feature logic.

Use Cases

01Architecting complex multi-screen flows with intricate state management dependencies.

02Defining new data models and persistence strategies for mobile or web applications.

03Gathering requirements for features with high 'blast radius' or non-obvious edge cases.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add willsigmon/sigstack deep-interview

For use in Claude.ai and ChatGPT

Download Skill