AI Evaluation
Prompt robustness, hallucination audits and reproducible verdicts
Evidence-driven AI evaluation, multi-turn interaction cases, and logic-boundary testing.
This portfolio is an evidence log of AI collaboration: what we shipped, how we tested it, and where it broke. Working alongside my AI partners Paw (小爪, validator) and Kyle (小克, builder), I design interaction cases and evaluation reports for multi-turn logic, information gaps and boundary failures.
Start with the evidence: evaluation reports · interaction cases · shipped work.
Crystaria (Boss) 👑
↓
Paw (Validator) 🦞
↓
Kyle (Builder) 👨💻
| Role | Responsibility | Focus |
|---|---|---|
| Crystaria | Decision, Communication | Strategy & Direction |
| Paw | Verification, Web Validation | Quality Assurance |
| Kyle | Implementation, Code/Git | Building & Execution |
| Category | Projects | Description |
|---|---|---|
| 🚀 Independent Apps | 1 | Full-stack applications with live demos |
| 🤖 ClawHub Skills | 6 | MCP tools published on ClawHub |
| 🤝 Community Contributions | 1 | Open-source project contributions |
| ✨ Experience Optimizers | 1 | Tools and scripts for better UX |
| Total | 9 |
A full-stack language learning and analysis platform built with React and Node.js.
Your extension has been successfully merged into the official SillyTavern-Content repository (PR #43), becoming part of the global SillyTavern ecosystem used by thousands of users worldwide.
| Skill | Downloads | Link |
|---|---|---|
| Loading data... | ||
Last updated: April 7, 2026