Does it work with my specific tech stack?

Yes, it verifies compliance by reading your project's local documentation (like CLAUDE.md) and searching for existing patterns within your specific repository.

How does the 90% confidence threshold work?

The skill evaluates five criteria (Duplicates, Architecture, Docs, OSS, Root Cause). If the weighted score is below 90%, it stops the implementation to request more context or clarification.

How much token cost can this skill save?

By spending roughly 100-200 tokens on a confidence check, users can avoid wasting 5,000 to 50,000 tokens on incorrect or hallucinated implementation attempts.

What is the Confidence Check skill for Claude Code?

It is a specialized skill that performs a pre-implementation audit to ensure a task has sufficient context, documentation, and architectural alignment before writing code.

Confidence Check

Name: Confidence Check
Author: caiomioto2

bycaiomioto2

•

Security & Testing

Assesses implementation readiness through a multi-factor 90% confidence threshold to prevent architectural drift and redundant code.

The Confidence Check skill is a rigorous pre-implementation safeguard designed to ensure AI-driven development remains accurate, efficient, and aligned with your project's goals. By evaluating five critical pillars—duplicate code detection, architectural alignment, official documentation verification, open-source reference analysis, and root-cause identification—it calculates a numerical confidence score. This prevents costly 'wrong-direction' work, saves significant token usage, and ensures that every code change adheres to established project patterns and official best practices before execution begins.

Key Features

01Automated codebase scanning to prevent duplicate implementations

02Architecture compliance verification against CLAUDE.md and PLANNING.md

03Root cause identification to ensure symptoms aren't confused with problems

04Weighted confidence scoring (0.0 - 1.0) with a 90% success threshold

051 GitHub stars

06Integration with official documentation and OSS reference lookups

Use Cases

01Ensuring new features align with specific tech stacks like Supabase or UV

02Reducing token costs by identifying flawed implementation plans early

03Preventing code duplication in large, complex codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add caiomioto2/superclaude_marketplaceplugin confidence-check

For use in Claude.ai and ChatGPT

Download Skill