I build with AI agents and design for humans to stay in control.
I build practical tools and workflows around AI agents, especially context engineering, human-in-the-loop automation, and systems that make AI-assisted work more reliable and reusable.
My projects begin with problems I encounter in my own work. I use coding agents to turn ideas into working software, while taking responsibility for product decisions, architecture, review, testing, and releases.
| Area | What I'm working on |
|---|---|
| Context Engineering | Project context that people and coding agents can understand, review, and reuse |
| Agent Workflows | Repeatable ways to move from an idea to implementation and verification |
| Local-first Tools | Applications that help people organize their work and choose a concrete next step |
A local context compiler for ChatGPT planning and Codex development.
Turns explicitly selected project documents and reviewed state into auditable Markdown, without model API calls or whole-repository uploads. I created the project and maintain its code, tests, documentation, and releases.
English guide · Releases · Contributing
A local-first personal direction and action system.
Helps people choose what matters now and turn it into a manageable next step. Progress stays on the user's own machine; the app works without an account or a runtime AI service.
Problem → system design → AI-assisted implementation → verification → real use → iteration
I care about:
- Context that stays understandable and traceable to its source.
- Automation with visible failures and clear privacy boundaries.
- Tools that fit everyday use and keep important decisions with people.
AI Context Linker has received an external contribution improving state diffs and provenance. Reviewing contributions and strengthening regression coverage are part of how I maintain it.
Issues, practical feedback, and focused pull requests are welcome. Reach me through the project repositories on GitHub.