Founder · Author · Maintainer
A personal, open-source Model Context Protocol server that gives AI agents secure, structured access to Microsoft Dataverse: schema, metadata, data, and platform operations through the Dataverse Web API.
- Designed and shipped a production-grade MCP server in TypeScript with about 80 structured tools across 28 categories, covering CRUD, OData and FetchXML queries, metadata, batch operations, RBAC, solutions, audit, and schema writes.
- Implemented three enterprise authentication modes (device code, client credentials, managed identity via MSAL) with AES-256-GCM encrypted token caching and guardrails on destructive operations.
- Distributed via npm/npx (mcp-dataverse, about 900 monthly downloads) and listed on the Smithery and Glama MCP registries. Works with Claude, VS Code Copilot, Cursor, Windsurf, and any MCP client.
- Production-quality engineering: modular architecture, structured outputs, stdio and HTTP/SSE transports, automated tests, CI, a documentation site, a changelog, and a public roadmap.
Graph Orchestrator
Creator
A personal project, and also how I run my own AI-assisted work. A portable multi-project orchestrator that breaks a goal into ordered tasks and drives each one through a plan, build, and review loop, backed by a Neo4j graph database.
- Built a harness that runs each project's own verification command after every build and stores the result as evidence, so an agent's self-report never decides whether a build passed.
- Modeled projects, tasks, phases, and evidence as a graph, with a lessons store that carries fixes and decisions forward into new projects and fresh agent sessions.
- Use it daily to plan, build, and review my own work, including the Agentic Delivery Framework below.
Agentic Delivery Framework
Creator · Lead
A delivery framework that structures the full software delivery lifecycle: discovery, architecture, implementation, review, and production deployment, for AI-assisted teams. Started as a personal initiative. My practice director sponsored it, and Hitachi Solutions adopted it as internal standard practice.
- Enforces human-in-the-loop checkpoints at architecture decisions, implementation gates, and ALM promotion, giving production governance to agentic delivery at scale.
- Runtime integrates MCP servers to ground recommendations in live project context. Installable into any target project via CLI.
- Started as a personal initiative. My practice director sponsored it, and it rolled out across teams and business domains as standard delivery practice.
AI-Assisted Dev Context Manager
Creator
Personal project. A context and memory tool for AI-assisted development across large, multi-team codebases. It keeps track of repository knowledge such as architecture decisions, patterns, and team conventions, so coding agents can pull in the right context automatically.
- Structures repository knowledge so AI coding tools can retrieve relevant context automatically across sessions.
- Uses vector search and Azure-native services to support retrieval across large, multi-team projects.
- An exploratory, personal-use project. Not deployed as an official internal tool.
