I build practical AI tools that keep context, operate real workflows, and turn rough ideas into usable systems.
English | 中文
Current Projects | Focus Areas | Skills | GitHub Snapshot | Connect
- Project Brain Studio - Local memory dashboard and Codex skill package for project-specific context, MCP tools, checkpoints, and memory review.
- github-projects-and-articles - Workspace for GitHub project tracking, writing, and public technical notes.
- infinite-canvas - AI creative canvas for image/video generation, prompt workflows, reference editing, and multi-agent creation.
- ppt-master - AI-driven SVG and presentation generation system.
- v2apiCodeHub - TypeScript tooling and API workspace for building and organizing development utilities.
- opencli - Universal CLI hub concept for making tools discoverable and executable by agents.
Building Project Brain as a local memory layer for Codex
Exploring agent-friendly interfaces, MCP tools, and workflow automation
Improving personal GitHub projects into clean, public, trackable repos
|
Project-aware context, checkpoints, searchable conversation references, and reviewable long-term memories. |
Small tools that help agents inspect projects, operate workflows, and make useful decisions with less repeated setup. |
|
Canvas, image, video, prompt, and multi-agent creation flows where people keep control over the process. |
Practical scripts, dashboards, and integrations that reduce manual work without hiding important state. |
- Tools should be useful in messy real projects, not only in demos.
- AI agents need memory, context boundaries, and reviewable decisions.
- Good automation should reduce repeated work without hiding important state.
- Creative AI tools should give humans more control, not less.
Simple tools. Reviewable context. Better workflows.
