doubt-driven-development

Review high-stakes decisions through a structured adversarial CLAIM, EXTRACT, DOUBT, RECONCILE, and STOP process.

31|20|Updated Apr 9, 2026
One-click install
npx skills add https://github.com/mycelium-hq/ai-brain-starter --skill doubt-driven-development-mycelium-hq
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doubt-driven-development
Source: https://github.com/mycelium-hq/ai-brain-starter/tree/main/skills/doubt-driven-development
Command: npx skills add https://github.com/mycelium-hq/ai-brain-starter --skill doubt-driven-development-mycelium-hq

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Non-trivial decisions are prone to drift as context accumulates, leading to overconfident yet incorrect outputs. This skill injects a fresh-context adversarial reviewer to surface hidden assumptions and counterarguments before finalizing decisions.

Core Features & Use Cases

  • Fresh-context adversarial review before outputs for high-stakes decisions.
  • Structured workflow: CLAIM → EXTRACT → DOUBT → RECONCILE → STOP with optional cross-model escalation.
  • Clear loading constraints and safeguards to avoid nested doubt or unsafe automation on critical code paths.

Quick Start

Use this skill to instigate a fresh-context adversarial review for high-stakes decisions in your main session.

Frequently Asked Questions about doubt-driven-development

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

FAQPage Schema
What is fresh-context adversarial review for code decisions?

Fresh-context adversarial review is a validation technique that surfaces hidden assumptions and counterarguments in non-trivial architectural choices before finalizing high-stakes decisions. It prevents context drift by forcing a structured doubt process to challenge accumulated conclusions.

How do I validate architectural choices in an unfamiliar codebase?

You validate architectural choices by applying a structured CLAIM, EXTRACT, DOUBT, RECONCILE, and STOP workflow. This process forces an adversarial review of risky code changes and irreversible decisions to identify potential issues before any output stands.

When should I use a doubt-driven approach for decision-making?

Use a doubt-driven approach for high-stakes decision-making involving irreversible impact, risky code changes, or architectural choices across unfamiliar codebases. It is essential when accumulated context leads to overconfident yet incorrect outputs.

Can I use cross-model review to validate risky code changes?

Yes, you can use optional cross-model review to escalate the validation of risky code changes. This injects a fresh-context adversarial reviewer to surface counterarguments and hidden assumptions before finalizing critical decisions.

What are the limitations of using adversarial review on critical code paths?

Limitations include the need for strict sandbox safety to avoid unsafe automation on critical code paths, and clear loading constraints to prevent nested doubt loops from interfering with the adversarial review process.