Can it prevent me from creating duplicate database models?

Yes. The skill includes workflows to scan your Prisma schemas and TypeScript interfaces to ensure you aren't creating a new model for an entity that already exists.

What specifically does this skill detect as a hallucination?

It detects 'technical hallucinations' such as suggesting React for a Vue project, proposing non-existent npm packages, inventing API endpoints that don't exist, or suggesting database functions incompatible with your current version.

How does the detector know which technologies I use?

It uses your project's 'docs/project/tech-stack.md' file as the single source of truth to validate all framework, library, and infrastructure choices.

Does this work with external APIs like Stripe or AWS?

Absolutely. It verifies if external services are configured in your environment variables and checks suggested endpoints against documented API structures to prevent 'fake' API calls.

Hallucination Detector

Name: Hallucination Detector
Author: marcusgoll

bymarcusgoll

•

개발자 도구

Ensures technical accuracy by validating code suggestions against your project's documented tech stack and architecture.

The Hallucination Detector serves as a real-time technical auditor for AI-driven development, preventing the common pitfall of incorrect or invented technical decisions. It acts as a reality checker by cross-referencing all AI suggestions—including frameworks, npm packages, database schemas, and API endpoints—against your project's 'single source of truth' (tech-stack.md). By requiring evidence for technical claims and blocking architectural violations, it ensures that every implementation aligns perfectly with your existing codebase and prevents the introduction of redundant or incompatible technologies.

주요 기능

01Validates technical suggestions against docs/project/tech-stack.md

02Detects non-existent npm packages and fake external API endpoints

03Prevents duplicate entity creation by scanning existing database schemas

04Verifies library version compatibility and project-specific conventions

05Requires citations and evidence for all technical claims and patterns

0618 GitHub stars

사용 사례

01Maintaining consistency in large codebases by preventing redundant library additions

02Onboarding developers and AI agents to complex projects with strict technical guardrails

03Ensuring new feature implementations strictly adhere to the pre-defined project architecture

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add marcusgoll/spec-flow hallucination-detector

For use in Claude.ai and ChatGPT

Download Skill