doubt-driven-development

Runs adversarial fresh-context reviews of non-trivial outputs using doubt cycle CLAIM-EXTRACT-DOUBT-RECONCILE-STOP.

1|1|Updated Apr 11, 2026
One-click install
npx skills add https://github.com/Borisserz/FoodTracker --skill doubt-driven-development-borisserz
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doubt-driven-development
Source: https://github.com/Borisserz/FoodTracker/tree/main/.agents/skills/skills/doubt-driven-development
Command: npx skills add https://github.com/Borisserz/FoodTracker --skill doubt-driven-development-borisserz

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

A confident answer is not a correct one. Long sessions accumulate context that quietly turns assumptions into "facts" without anyone noticing. Doubt-driven development is the discipline of materializing a fresh-context reviewer — biased to disprove, not approve — before any non-trivial output stands.

Core Features & Use Cases

  • Fresh-context reviewer that challenges assumptions and surfaces blind spots.
  • Adversarial prompts and artifact-contract framing to test robustness.
  • A formal doubt cycle (CLAIM, EXTRACT, DOUBT, RECONCILE, STOP) to maintain auditable decisions.
  • When to use: for architectural decisions, production changes, security-sensitive logic, or any high-stakes code or data work.

Quick Start

Provide an artifact and a contract, and run an adversarial review to surface unstated assumptions and edge cases before proceeding.

Frequently Asked Questions about doubt-driven-development

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

FAQPage Schema
What is adversarial code review and how does it surface blind spots in architectural decisions?

Adversarial code review introduces a fresh-context reviewer biased toward disproof rather than approval. It challenges accumulated assumptions during high-stakes architectural decisions by materializing a reviewer focused on disproving outputs before they proceed to production.

How do I run an adversarial review for security-sensitive production changes?

Provide an artifact and a contract, then execute the structured doubt cycle: CLAIM, EXTRACT, DOUBT, RECONCILE, STOP. This formal process audits security-sensitive production changes by forcing disproof of assumptions before allowing the output to stand.

Can I use doubt-driven development for data workflows or is it limited to software engineering?

Doubt-driven development applies across both software and data workflows. It is designed for any non-trivial, high-stakes output where context is partial, enabling rigorous adversarial examination of decisions regardless of the specific engineering domain.

What's the difference between standard code review and a fresh-context adversarial review?

Standard code review typically seeks approval, while adversarial review uses a fresh-context reviewer explicitly biased to disprove. The structured doubt cycle maintains auditable decisions by testing robustness through adversarial prompts rather than validating existing assumptions.

When should I not use adversarial review for my code or data outputs?

Adversarial review targets non-trivial, high-stakes outputs like architectural decisions and security-sensitive logic. For routine or low-stakes changes where context is complete and consequences are minimal, the rigorous doubt cycle may introduce unnecessary overhead.

Does doubt-driven development support cross-model opinions to broaden code review coverage?

Yes, doubt-driven development supports optional cross-model opinions to broaden review coverage. This expands the adversarial examination by introducing diverse fresh-context perspectives to disprove assumptions in architectural and security-sensitive decisions.