Can I skip validation for simple tasks?

Yes, full validation can be skipped for single-file changes under 50 lines, documentation-only changes, or basic formatting updates.

How is the confidence score calculated?

The score is a 100-point total weighted across Requirements Completeness (25), Step Quality (25), Context Coverage (25), and Risk Assessment (25).

Does this skill help with codebase consistency?

Yes, the Integration Challenge specifically evaluates whether the proposed plan respects existing patterns, error handling, and testing conventions in your codebase.

What is the Five Challenges framework?

It is a set of adversarial reviews designed to 'break' a plan by testing it against requirements, edge cases, simplicity, integration conflicts, and failure modes.

What happens if a plan fails validation?

The skill initiates a revision loop to address the flaws. If the plan still fails after two loops, the issues are surfaced to the user for a final decision.

Plan Validation & Scoring

Name: Plan Validation & Scoring
Author: twofoldtech-dakota

bytwofoldtech-dakota

0•

安全与测试

Validates AI-generated development plans through adversarial challenges and quantitative confidence scoring to ensure production readiness.

The Validation skill provides a robust framework for auditing AI development plans before execution, preventing costly errors in complex codebases. It employs a two-tier approach: an adversarial 'Challenge Review' that rigorously tests plans against edge cases, simplicity, and integration requirements, followed by a 'Confidence Scoring' system that quantifies readiness across four critical dimensions. This ensures that every proposed change is atomic, verifiable, and aligned with existing architectural patterns before a single line of code is written, effectively acting as a pre-execution quality gate.

主要功能

010 GitHub stars

02Automated revision loops for plans that fall below specific confidence thresholds

03Context-aware assessment that integrates memory rules and discovered codebase patterns

04Adversarial 'Five Challenges' framework covering requirements, edge cases, simplicity, integration, and failure modes

05Quantitative Confidence Scoring on a 100-point scale across four dimensions

06Visual validation reporting with detailed ASCII status boards and verdicts

使用场景

01Pre-flight checking complex multi-step refactorings to prevent architectural regression

02Ensuring AI-generated plans align with specific team conventions and security constraints

03Validating edge-case handling for mission-critical features before deployment

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add twofoldtech-dakota/studio validation

For use in Claude.ai and ChatGPT

Download Skill