What problem does it solve?
Plans, designs, and architectural decisions often contain hidden assumptions that only surface after code is written. This Skill stress-tests thinking through a structured interview process, mapping every decision as a design tree so nothing is silently assumed before implementation begins.
Core Features & Use Cases
- Round-based frontier interviewing: Maps work as a design tree and asks every currently-unblocked decision in numbered rounds, each with a recommended answer.
- Fact gathering from the codebase: Reads config files, models, routes, CLI help, and CONTEXT.md in Stacks projects so users only answer decisions, never lookups.
- Shared interview primitive: Used directly or embedded by other workflows like plan review, product ideation, and redesign sessions.
- Use Case: Before building a new feature, run a grilling session to settle questions about data models, API surface, and configuration drivers, then hand the settled understanding to a planning step.
Quick Start
Ask the assistant to grill you on your feature plan or design decision before any code is written.