Ricko Yu
Continuously exploring efficient collaboration between humans and AI by building high-quality, local-first developer tools and agent harnesses.
At a Glance
- Go developers
- AI agent researchers
- Software engineers working with LLMs
AI Tools by Ricko Yu
(1)
mindwalk
CLI Agent Session Visualizer
Discussions
No discussions yet
Be the first to start a discussion about Ricko Yu
Latest News
Release of Mindwalk: Replaying coding-agent sessions as a 'Code City' map.
Introduction of Jeju Agent Harness: Design thinking and future vision.
New Blog Post: 'Making the Self-Evolution Loop Solid: From Bad Case to Harness Patch'.
Article: 'Don’t Teach Top-Tier Models to Think' - A shift in LLM management philosophy.
Products & Services
A live reload utility designed for Go application development, helping developers see changes instantly without manual restarts.
A visualization tool that maps coding-agent trajectories onto a 3D city-like representation of a codebase to aid in debugging and LLM evaluation.
A declarative, local-first runtime for bounded AI agents that allows developers to define agent behavior in a manifest and audit every effect.
Market Position
Differentiates by focusing on 'careful' toolcraft and high-fidelity visualization of agent behavior, moving beyond simple text logs to structured, 3D spatial representations of codebase interactions.
Leadership
Founders
Ricko Yu
AI Engineer and open-source developer based in Shanghai and Suzhou. Creator of high-impact developer tools including 'Air' for Go live reloading and 'Mindwalk' for AI agent visualization. Formerly a backend engineer with experience in Golang, Python, and AI agent development.
Executive Team
Ricko Yu
Solo Developer / AI Engineer
Expert in AI-human collaboration and Go ecosystem. Focuses on building observability and control harnesses for LLM agents.
Founding Story
Ricko Yu started by creating tools to solve personal pain points in the Go development lifecycle (Air). Recently, his focus shifted to the 'Agentic AI' era, creating Mindwalk and Jeju to improve the observability and safety of AI agents.
Business Model
Revenue Model
Primarily open-source development (MIT licensed). Ricko Yu also utilizes affiliate marketing for developer services (e.g., BandwagonHost, Aliyun) on his personal blog.
Pricing Tiers
Core tools like Air, Mindwalk, and Jeju are free to use under the MIT license.
Target Markets
- Go developers
- AI agent researchers
- Software engineers working with LLMs
- Streamlining Go web development
- Visualizing LLM coding agent trajectories
- Evaluating LLM task performance in complex repos
- Building safe and bounded AI agents