claude-authenticity

Name: claude-authenticity
Availability: InStock
Author: AgentScope-AI

Verify Claude API endpoint authenticity and extract injected system prompts.

775|63|Updated Jul 8, 2025

Search Tags:#system prompt #claude #authenticity #endpoint testing #llm security #api verification

Ready to run

npx skills add https://github.com/agentscope-ai/OpenJudge --skill claude-authenticity

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: claude-authenticity
Source: https://github.com/agentscope-ai/OpenJudge/tree/main/skills/claude-authenticity
Command: npx skills add https://github.com/agentscope-ai/OpenJudge --skill claude-authenticity

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires httpx.

What problem does it solve?

This Skill helps you determine if an API endpoint is genuinely powered by Anthropic's Claude models, protecting against impersonators and wrappers. It also identifies if a custom system prompt has been injected.

Core Features & Use Cases

Authenticity Verification: Uses 9 weighted checks to score the likelihood of an endpoint being genuine Claude.
System Prompt Detection: Extracts any injected system prompts that might override Claude's default behavior.
Use Case: You're evaluating a new AI service that claims to use Claude. Run this Skill to confirm its authenticity and check for any hidden instructions before integrating it into your workflow.

Quick Start

Use the claude-authenticity skill to verify the API endpoint 'https://your-provider.com/v1/messages' with model 'claude-opus-4-6' and API key 'sk-xxx'.

Frequently Asked Questions about claude-authenticity

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

FAQPage Schema

How do I verify if an API endpoint is genuinely powered by Claude?▼

To verify Claude API authenticity, this Skill runs 9 weighted rule-based checks against the endpoint to score its legitimacy, detecting wrappers and impersonators claiming to use Claude models.

Can I detect injected system prompts when testing an API endpoint?▼

Yes, system prompt detection extracts any injected custom instructions from the endpoint response, revealing hidden prompts that might override Claude's default behavior before you integrate the service.

Does Claude authenticity verification work with OpenAI-compatible API formats?▼

Claude authenticity verification supports both Anthropic-native and OpenAI-compatible API formats, with a preference for the native format because it provides a richer response structure for analysis.

What's the best way to check a new AI service for Claude impersonators before integration?▼

The best way to check for Claude impersonators is running weighted endpoint testing that scores authenticity across 9 checks, extracting any injected system prompts to identify wrappers misrepresenting their underlying models.

Why does LLM security testing require endpoint verification for Claude APIs?▼

LLM security testing requires endpoint verification because API wrappers and impersonators can misrepresent their underlying models, making authenticity checks necessary to confirm genuine Claude usage and expose hidden system prompts.

Do I need an API key and model name to verify Claude endpoint authenticity?▼

Yes, you need the provider's API endpoint URL, the target model name such as claude-opus-4-6, and your API key to authenticate the request and run the weighted authenticity checks.