What is the 90% confidence threshold?

It is a mandatory readiness score calculated across five checks; if the combined score is below 0.90, the skill stops implementation and requests more context to prevent errors.

Can it detect existing duplicate code?

Yes, it utilizes grep and glob commands to search your codebase for similar functions or modules before creating new ones.

Which documentation sources does it verify?

The skill uses MCP tools like Context7 and WebFetch to cross-reference your implementation plan with official documentation and proven open-source examples.

How does this skill save AI tokens?

By investing approximately 100-200 tokens in a preliminary check, it prevents the AI from generating thousands of tokens of incorrect, redundant, or non-compliant code.

Confidence Check

Name: Confidence Check
Author: SuperClaude-Org

bySuperClaude-Org

•

19,773

•

生产力与工作流

Validates implementation readiness through a rigorous five-point confidence assessment to prevent wasted tokens and logical errors.

The Confidence Check skill is a strategic decision-gate utility for Claude Code designed to minimize 'wrong-direction' development by ensuring all prerequisites for a task are met before implementation begins. By evaluating five critical criteria—lack of duplicates, architectural alignment, official documentation verification, open-source references, and root cause clarity—it generates a mathematical confidence score. This systematic approach significantly reduces token waste and improves project success rates by enforcing a 90% confidence threshold before any code modifications are permitted.

主要功能

01Integration with MCP tools for official documentation and OSS reference lookups

02Multi-factor readiness scoring based on five distinct architectural and logic checks

03Tech stack alignment verification against project-specific patterns and planning files

0419,773 GitHub stars

05High-ROI token management by preventing extensive refactoring of incorrect solutions

06Automated codebase scanning to prevent duplicate implementations of existing logic

使用场景

01Resource optimization for large-scale projects where token costs are a primary concern

02Pre-implementation validation for complex features to ensure architectural compliance

03Bug fix verification to ensure the root cause is identified before writing code

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add superclaude-org/superclaude_framework confidence-check

For use in Claude.ai and ChatGPT

Download Skill