doubt-driven-development

Apply bounded adversarial review to surface counter-evidence for high-risk decisions.

Updated Jun 4, 2026
One-click install
npx skills add https://github.com/zxyycom/docnav --skill doubt-driven-development-zxyycom
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: doubt-driven-development
Source: https://github.com/zxyycom/docnav/tree/main/.codex/skills/doubt-driven-development
Command: npx skills add https://github.com/zxyycom/docnav --skill doubt-driven-development-zxyycom

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Doubt-driven development enforces a fresh-context adversarial review before committing non-trivial decisions, preventing overconfidence and catching edge cases early.

Core Features & Use Cases

  • Structured Claim-Contract-Evidence workflow to surface gaps or issues before implementation.
  • Bounded adversarial review process with steps (CLAIM, CONTRACT, ARTIFACT, DOUBT, RECONCILE, STOP) and optional cross-model escalation.
  • Fresh-context reviewer integration for high-stakes decisions across architectures, security-sensitive logic, migrations, and compatibility scenarios.

Quick Start

Initiate a doubt-driven cycle by submitting the artifact and contract to an adversarial reviewer to surface concrete findings.

Frequently Asked Questions about doubt-driven-development

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

FAQPage Schema
How do I perform an adversarial review for high-stakes architectural decisions?

Adversarial review for architectural decisions is performed by submitting the artifact and contract to a fresh-context reviewer to surface concrete counter-evidence before finalizing the claim. This bounded process prevents overconfidence and catches edge cases early.

What is doubt-driven development and when should I use it for migrations?

Doubt-driven development is a structured workflow enforcing adversarial review before committing non-trivial decisions. Use it for migrations, security-sensitive logic, or cross-service contracts to surface gaps and issues before implementation begins.

How do I apply a claim-contract-evidence workflow to validate cross-service contracts?

To validate cross-service contracts, apply the CLAIM, CONTRACT, ARTIFACT, DOUBT, RECONCILE, and STOP steps to classify artifacts and surface counter-evidence. This structured workflow identifies gaps or issues before implementation finalization.

Can I use a fresh-context reviewer to catch edge cases in security-sensitive logic?

Yes, a fresh-context reviewer can be used to catch edge cases in security-sensitive logic through a bounded adversarial review. The process surfaces concrete findings by applying the claim-contract-evidence workflow with clear stop conditions.

What are the limitations of bounded adversarial review for correctness-critical decisions?

Bounded adversarial review for correctness-critical decisions is limited by its defined stop conditions and artifact classification scope. It focuses on surfacing counter-evidence before finalizing a claim, relying on optional cross-model escalation rather than exhaustive manual analysis.

Why does my architectural review miss edge cases before implementation?

Architectural review misses edge cases when overconfidence prevents fresh-context analysis. Applying a structured claim-contract-evidence workflow with adversarial review surfaces counter-evidence and gaps before finalizing non-trivial decisions.