I build reliable AI-agent skills, research workflows, and practical automation for Claude Code, OpenAI Codex, OpenCode, ChatGPT, and Windows.
My projects focus on one thing: turning complex, fragile workflows into tools that are easier to discover, verify, install, continue, and share.
Browse 588 AI agent skills · See all repositories · Latest Windows tool
| Project | What it solves |
|---|---|
| Universal AI Skill Library | 588 searchable, bilingual agent skills mapped across Claude Code, Codex, and OpenCode, with direct downloads and compatibility notes. |
| Super Memory | A file-based project memory layer for long sessions, context-window handoffs, and switching between coding agents. |
| Map Skill | Evidence-aware multimodal route planning with a universal edition plus Istanbul and 15 major-city editions. |
| BetterDiscord Auto Repair Launcher | A short-lived Windows launcher that repairs BetterDiscord after Discord updates without scheduled tasks, polling, or VBS. |
| Universal AI Finance Skills | Evidence-based investing, financial literacy, BIST/TEFAS, portfolio risk, warrants, and calibrated forecasting skills. |
| Scientific Agent Skills | Reproducible scientific-research skills spanning bioinformatics, single-cell analysis, Nextflow, R, and data quality. |
Also: Truth-First Counterargument, an evidence-first workflow that verifies claims, concedes when warranted, and builds sourced rebuttals.
- Evidence first: claims, integrations, and release artifacts are verified before I call them complete.
- Beginner-friendly delivery: English/Türkçe documentation, direct downloads, clear install paths, and practical troubleshooting.
- Safe automation: bounded retries, narrow permissions, backups, checksums, and explicit limits instead of silent overreach.
- Portable workflows: tools designed to move between agents and survive long-running project handoffs.
- reusable AI-agent skill libraries and discovery
- context engineering and project memory
- research workflows with honest uncertainty
- small Windows tools that remove repetitive manual fixes
If a project is useful, opening an issue with a real use case or sharing it with the right community helps more than a generic star request.