I build AI tools for coding, learning, and creative work. My current focus is on coding agents, voice interfaces, and local-first workflows — making useful AI capabilities easier to work with.

I work at the intersection of software engineering and human-computer interaction: turning model capabilities into tools with clear workflows, useful feedback, and room for human judgment.

Website · Email · Explore projects · All repositories

Ayatori Nexus / CLI Model Switcher	Vibe Coding Voice
Project demo page	Windows app preview
For working across several coding agents without rebuilding your setup each time. Switch agent and provider profiles. Reuse shared context and handoff templates. Check configuration and move portable state between machines. Source & setup · Releases	For speaking through prompts and code changes, especially Chinese and mixed-language dictation. Start and stop recording with a global shortcut. Choose local or online transcription. Edit, copy, or send the resulting text. Alpha downloads · Model setup

_{Local Codex activity · Costs are API-equivalent estimates, may exclude unpriced models, and are not subscription charges.}

• DOM Native PPTX — convert slide-safe HTML into native, editable PowerPoint objects, with explicit validation and raster policies. Beta. Quick start.
• XS Skill — a Chinese short-fiction writing skill for story planning, revision, and continuity checks. Writing methods.
• XHS AI Content Publisher — turn public sources and supplied notes into source-linked Xiaohongshu and WeChat drafts, with review and visual-card planning.
• AI Comic — a prototype connecting scripts, storyboards, character cards, and comic layouts.

• Fairy DSH for Windows — a Windows adaptation of Fairy DSH, with an isolated runtime, browser tools, and local GPT-SoVITS voice.
• Bilibili–Obsidian Clipper — a fork connecting video subtitles with Markdown and Obsidian; part of my exploration of local knowledge workflows.

• Carbon Clash — a civic strategy game exploring environmental trade-offs, with optional AI integration.

• Clear agent interactions. Make task state, progress, errors, and recovery understandable.
• Repeatable workflows. Turn useful experiments into tools that can be configured, reused, and maintained.
• Local ownership. Keep data portable and make the choice between local and online processing explicit.
• Human control. Give people useful feedback and a way to inspect, edit, or override the result.

• MSc Human-Computer Interaction — Newcastle University
• BSc Computer Science — Heriot-Watt University

I work with Python, TypeScript, React, Node.js, Rust, Tauri, and Shell, across CLI utilities, desktop apps, browser tools, and AI-assisted creative workflows.

For project questions, open an issue in the relevant repository. For a conversation about AI tools, agent UX, or creative workflows, send me an email.

Personal profile card