Skip to content
View superWorldSavior's full-sized avatar

Block or report superWorldSavior

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
superWorldSavior/README.md

Systems that let agents do real work: intent, model, run, trace, evidence

Hi, I'm Erwan.

I think in systems, build in code, and teach from the work.

I'm a software architect and the founder of Casys, based in Taiwan. I work where agent protocols, engineering models, business systems, data, and decision-making meet.

I like turning complicated real-world work into software that is executable, inspectable, and useful to the people who own it.

I develop tools and software on my personal GitHub. Existing npm and JSR packages keep the @casys namespace.

Through Casys — Complex Adaptive Systems, I contribute to the discipline of Agentic Process Engineering (APE) and offer consulting and education on the design and governance of processes with AI. Casys publishes the APE handbook and its supporting resources.

I provide no-code training through THE NO CODE GUY, my legal company, which operates the Casys brand.

Open source · APE book · Field notes · Teaching · LinkedIn

What I build

  • Agent infrastructure — MCP systems and operator surfaces that connect agents to the tools, permissions, knowledge, and traces behind real workflows.
  • Engineering software — from system architecture to CAD and simulation, with evidence that carries into the wider workflow.
  • Data, graphs, and optimisation — knowledge structures, exact models, heuristics, and interfaces that make difficult decisions legible.

My favourite loop

understand → model → build → run → document → teach

I publish field notes, diagrams, and working examples alongside the code. Documentation is part of the product, and teaching is how I test whether the ideas remain useful outside the repository where they started.

Selected work

My tools and software live on my personal GitHub. These projects capture the range:

  • mcp-platform — a TypeScript framework for MCP servers, apps, and composition.
  • casys-trader — experimental LLM-assisted paper trading with explicit risk controls and traceable decisions.
  • casys-digital-thread — experimental engineering workflows from reviewed intent to traceable artifacts and evidence.
  • mcp-erpnext — agent-operated ERPNext documents and workflows with interactive views.
  • editorial-workbench — a local editorial reader for HTML manuscripts and detailed Markdown outlines.

Side quest

NutrientCompare is a transparent supplement comparison project built for Taiwan.

If you're building a system where an agent needs to do more than talk, say hello.

Pinned Loading

  1. mcp-platform mcp-platform Public

    A Deno-first TypeScript framework for building Model Context Protocol (MCP) servers, compatible with Node.js.

    TypeScript 1

  2. casys-trader casys-trader Public

    LLM-assisted paper-trading daemon with risk controls and auditable simulated execution.

    Python

  3. casys-digital-thread casys-digital-thread Public

    Let AI agents take a design from brief to proof — does it behave, can we make it, what does it cost

    TypeScript 1

  4. mcp-erpnext mcp-erpnext Public

    Let AI agents operate your ERPNext — documents, workflows, and interactive viewers

    TypeScript 104 40

  5. editorial-workbench editorial-workbench Public

    Local-first editorial reader for an HTML manuscript and its detailed Markdown outline

    JavaScript