rt22766rt22766Communityยท1 Agent Skills Included

claude-skill-model-fingerprint

Verify model authenticity with fingerprint and stress tests

Detects whether a running model is a genuine Claude by checking reasoning effort fingerprints, magic string refusals, and self-reported identity. Runs parallel reasoning, date memory, and knowledge stress tests to expose downgraded, repackaged, or relayed models. Escalates to deep review with math reasoning and quote-output tests, then produces a structured authenticity report.
npx skills add rt22766/claude-skill-model-fingerprint --all -g -y

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install claude-skill-model-fingerprint?โ–ผ

Run `npx skills add rt22766/claude-skill-model-fingerprint --all -g -y` in your terminal to install the skill globally.

How to check if my Claude API is real?โ–ผ

Ask your agent to run the model fingerprint detection skill. It tests reasoning effort values, magic string refusals, and self-reported identity to flag fake or relayed models.

What tests does this skill run?โ–ผ

It runs parallel core tests including a candy logic puzzle, offline date recall, and anime knowledge questions, plus optional deep tests like a hard math sequence problem and a curly-quote output check.

Are the detection results officially certified?โ–ผ

No. The tests are community-sourced heuristics, not an Anthropic certification. You should combine multiple signals instead of trusting any single result.

Does it work with Claude Code?โ–ผ

Yes. It is built for Claude Code and runs its core tests through three parallel subagents to finish quickly.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’