개인 개발자이자, 동시에 AI Agent OS Lab이라는 감각으로 프로젝트를 만들고 있습니다.
I build practical things around AI agents, MCP, research workflows, and developer tooling.

쉽게 말하면,

• 새로운 AI workflow를 실험하고
• 쓸만한 구조로 정리하고
• 실제 도구와 프로젝트로 만드는 사람입니다

• AI Agents
• MCP tooling
• Research automation
• Agentic workflows
• Developer tools for the AI-native era

AI Agent OS Lab는 거창한 기관 이름이라기보다,
제가 AI agent systems, workflows, interfaces, and applied experiments를 계속 쌓아가는 작업 공간에 가깝습니다.

여기서는 보통 이런 걸 합니다:

• ideas → prototypes
• prototypes → usable tools
• tools → repeatable workflows

복잡한 AI 흐름을 그냥 신기한 데모로 끝내지 않고,
people can actually use 하는 형태로 바꾸는 데 관심이 많습니다.

• clawfit
  Agent + LLM + Hardware recommendation engine and evidence hub

• hello-mcp
  Beginner-friendly MCP onboarding utility for real-world use

• plugin-for-claude-research
  A structured research framework for Claude Code

• claude-agents-power-mcp-server
  Project-specific role recommendation and agent setup via MCP

AI Agents AI Agent OS MCP Research Automation Agentic Workflows Developer Tooling Applied AI Experiments Builder

• AI Agent OS Lab
  projects, experiments, direction, and the broader story behind what I'm building

Turning complex AI workflows into practical tools, usable systems, and lab-style experiments.

One of my recent contributions is improving OpenClaw Talk Mode for CJK languages and real-world voice use.

Related PRs:

• #53553 — fix(talk): Talk Mode TTS improvements for CJK languages (open)
• #53511 — fix(talk): prevent double TTS playback when system voice times out (merged)

What this work focused on:

• better Talk Mode TTS behavior for CJK languages
• preventing double playback in edge cases
• making spoken interaction more stable and usable
• refining real-world voice UX instead of stopping at a demo