virtual-think-tank

Runs a structured multi-phase debate panel to frame decisions and draft ADRs.

1|Updated Jan 4, 2026
One-click install
npx skills add https://github.com/stefaniuk/loadout --skill virtual-think-tank-stefaniuk
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: virtual-think-tank
Source: https://github.com/stefaniuk/loadout/tree/main/.github/skills/virtual-think-tank
Command: npx skills add https://github.com/stefaniuk/loadout --skill virtual-think-tank-stefaniuk

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes assets (resource) components.

What problem does it solve? Complex engineering, architecture, product, and organisational decisions often get one-shot answers that hide trade-offs and assumptions. This Skill forces a disciplined decision workshop with conflicting viewpoints, explicit trade-offs, and a mandatory human decision instead of a single recommendation. ## Core Features & Use Cases - Five-phase decision workflow: Frame the decision, build a neutral baseline, assemble a panel, run a structured debate, then force an explicit human choice with a review date. - Three panel styles: Choose archetype voices for technical decisions, stakeholder voices for organisational and process decisions, or named public figures with confidence levels and verification paths. - Guided and full modes: Run phase-by-phase with checkpoints when the brief is rough, or run phases 1-4 in one pass when the brief is complete. - ADR drafting: Produces a first-pass Architecture Decision Record aligned with the repository ADR template, defaulting status to Proposed and marking gaps as TBD. - Use Case: Paste a brief about whether to migrate a modular monolith to microservices, and receive a moderated debate between opposing architecture voices, decisive questions, and a draft ADR once you choose an option. ## Quick Start Invoke the skill with /virtual-think-tank and paste a filled-in briefing template describing your decision subject, constraints, and decision horizon.

Frequently Asked Questions about virtual-think-tank

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

FAQPage Schema
How do I run a structured decision debate with an AI assistant?▼

Invoke the skill with /virtual-think-tank and paste a structured brief covering the decision subject, current state, constraints, and decision horizon. The skill runs five phases: framing, neutral baseline, panel building, debate, and a forced human decision checkpoint.

What is the difference between guided and full mode in a decision workshop?▼

Guided mode completes one phase at a time and stops after each phase to ask whether to continue, correct, or switch panel style. Full mode runs phases 1 through 4 in a single response but only when the brief is complete enough, then stops at the decision checkpoint.

When should I use stakeholder voices instead of named public figures?▼

Use stakeholder voices for organisational, commercial, or process decisions where internal roles matter more than public thought leaders. Use named figures only when real public viewpoints add value, since each named voice requires a confidence level and verification path.

Can the skill make the final decision for me?▼

No. The skill explicitly refuses to make the final decision and instead forces you to choose option A, B, C, or defer with a date. You must state why you chose it, what you are giving up, and what would reverse the decision.

What happens if my decision brief is incomplete?▼

If three or more critical fields are missing, the skill stops and asks focused clarifying questions. If only one or two non-critical fields are missing, it makes explicit assumptions marked as Assumption and continues.

Does the skill produce an Architecture Decision Record?▼

Yes. After you choose an option, it drafts a concise first-pass ADR using the repository ADR template, defaulting the status to Proposed and marking missing metadata as TBD rather than inventing facts.