Open source infrastructure @ IBM · IBM i · Agentic AI · MCP · LLMs
I build open source infrastructure that connects IBM i to the modern AI stack - MCP servers, agents, and the tooling that makes LLMs useful on real enterprise systems.
Most of my work sits at the intersection of:
- IBM i & Db2 for i - system services, SQL tooling, PASE/JVM runtime concerns
- Model Context Protocol (MCP) - stdio/HTTP transports, YAML-driven tools, security defaults
- Agentic AI - giving agents safe, observable access to live system data
- Developer experience - CLIs, smoke tests, docs, and contracts that hold across languages
A minimal Java MCP server for IBM i — a faithful subset of the reference Node implementation. Declarative YAML tools, Mapepire connectivity, tokenizer-based SQL security, opt-in execute_sql, and hot-reload.
Java 17 · MCP Java SDK · mapepire-java
Production-grade TypeScript MCP server + ibmi CLI for IBM i. YAML toolsets, HTTP auth, pagination, and the reference contract that lite implements against.
TypeScript · npm workspaces · @ibm/ibmi-mcp-server
Exploring how LLM agents can safely explore, query, and reason about IBM i systems through MCP — without sacrificing read-only defaults, auditability, or operator control.
- GitHub: @wyattmogelson
- LinkedIn: your profile
- IBM open source: IBM GitHub


