agentscope-ai avatar

Claude Authenticity

This is a self-contained Python script that runs nine weighted checks to detect whether an API endpoint actually serves genuine Claude or just wraps someone else's model. It mirrors the claude-verify project's methodology, looking for signature fields, extended thinking blocks, cache_creation tokens, and identity keywords that real Claude returns but wrappers typically strip or fake. The script also extracts injected system prompts if a provider is overriding Claude's identity. One sharp design choice: it defaults to Anthropic's native API format because OpenAI-compatible endpoints silently drop the telltale fields, causing false negatives. You get a score out of 100 and a verdict: genuine, suspected, or likely fake. Useful when auditing third-party Claude resellers or testing whether your API key points to the real thing.

install copies · 90 days. Times the install command was copied on this site in the last 90 days, bots filtered.
6
GitHub stars
657
last updated
Jun 9, 2026

Install

$ npx -y skills add agentscope-ai/openjudge --skill claude-authenticity --agent claude-code

Installs into .claude/skills of the current project.

Files

Claude Authenticity Skill

Verify whether an API endpoint serves genuine Claude and optionally extract any injected system prompt.

No installation required beyond httpx. Copy the code blocks below directly into a single .py file and run — no openjudge, no cookbooks, no other setup.

pip install httpx

The 9 checks (mirrors claude-verify)

#CheckWeightSignal
1Signature 长度12signature field in response (official API exclusive)
2身份回答12Reply mentions claude code / cli / command
3Thinking 输出14Extended-thinking block present
4Thinking 身份8Thinking text references Claude Code / CLI
5响应结构14id + cache_creation fields present
6系统提示词10No prompt-injection signals (reverse check)
7工具支持12Reply mentions bash / file / read / write
8多轮对话10Identity keywords appear ≥ 2 times
9Output Config10cache_creation or service_tier present

Score → verdict: ≥ 85 → genuine 正版 ✓ / 60–84 → suspected 疑似 ? / < 60 → likely_fake 非正版 ✗

Sponsors

Details

Repository
github.com/agentscope-ai/openjudge
License
Apache-2.0
First seen
Jun 11, 2026

More from agentscope-ai/openjudge

All 9

Recommended

1juliusbrussee avatarcavemanjuliusbrussee/cavemanUltra-compressed communication mode. Cuts output tokens 65% (measured) by speaking like caveman while keeping full technical accuracy. Supports intensity levels618.3Kskills.sh installs
2mattpocock avatargrill-memattpocock/skillsA relentless interview to sharpen a plan or design.1.5Mskills.sh installs
3shadcn avatarimproveshadcn/improveSurvey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on s40Kskills.sh installs
4obra avatarsystematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes326.7Kskills.sh installs
5forrestchang avatarkarpathy-guidelinesforrestchang/andrej-karpathy-skillsBehavioral guidelines to reduce common LLM coding mistakes through explicit assumptions, simplicity, and verifiable success criteria.17.9Kskills.sh installs
6vercel-labs avatarfind-skillsvercel-labs/skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express int4.3Mskills.sh installs