doubt-driven-development

Subjects non-trivial decisions to fresh-context adversarial review before they stand.

3|Updated Jul 28, 2026
One-click install
npx skills add https://github.com/marcmarti9/agentit --skill doubt-driven-development-marcmarti9
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: doubt-driven-development
Source: https://github.com/marcmarti9/agentit/tree/main/skills/doubt-driven-development
Command: npx skills add https://github.com/marcmarti9/agentit --skill doubt-driven-development-marcmarti9

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Confident answers are not always correct ones, and long sessions quietly turn assumptions into facts. This Skill prevents costly mistakes by forcing every non-trivial decision through an adversarial fresh-context review while course-correction is still cheap. ## Core Features & Use Cases - Five-Step Doubt Cycle: A bounded CLAIM, EXTRACT, DOUBT, RECONCILE, STOP loop that isolates the smallest reviewable artifact and contract, then cross-examines it with an adversarial prompt. - Cross-Model Escalation: Optionally routes the artifact and contract to external CLIs like Gemini or Codex in read-only sandboxes to catch single-model blind spots, with explicit user authorization each time. - Structured Reconciliation: Classifies every reviewer finding as contract misread, actionable, trade-off, or noise, with a hard stop after three cycles. - Use Case: Before committing a new caching layer claimed to be thread-safe, extract the diff and its contract, spawn a fresh-context reviewer instructed to find issues, reconcile the findings, and only then proceed. ## Quick Start Apply doubt-driven development to the architectural decision I am about to make and cross-examine it with a fresh-context adversarial review before I commit.

Frequently Asked Questions about doubt-driven-development

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

FAQPage Schema
How do I stress-test a code decision before committing it?▼

Use the doubt cycle: write a compact CLAIM, extract the smallest reviewable artifact plus its contract, then spawn a fresh-context reviewer with an adversarial prompt that finds issues rather than validating. Classify each finding and stop after trivial results or three cycles.

What is the difference between doubt-driven development and code review?▼

Doubt-driven development is an in-flight posture applied per decision while changes are still cheap, whereas code review is a post-hoc verdict on a finished artifact like a PR. The two are complementary and can be used together.

When should I not use adversarial decision review?▼

Skip it for mechanical operations like renaming or formatting, clear unambiguous instructions, reading existing code, one-line obvious changes, pure tooling operations, or when the user explicitly prioritizes speed over verification.

Can doubt-driven development run inside a subagent or persona?▼

No, it is designed for the main-session orchestrator because spawning a fresh-context reviewer from a persona violates orchestration rules. In nested contexts it degrades to a flagged self-questioning fallback or escalates to the main session.

How does cross-model review work with Gemini or Codex CLI?▼

After single-model review, the user is offered a cross-model second opinion. If accepted, the artifact and contract are piped via stdin to a verified CLI in a read-only sandbox, with the exact invocation confirmed and authorized by the user each time.

Why must the reviewer not receive the original claim or reasoning?▼

Passing the claim or conclusions biases the reviewer toward agreement. The reviewer receives only the artifact and contract so it independently determines whether the artifact satisfies the contract.