AI Charts
AI-powered flowchart, ERD, and swimlane diagram builder with a built-in AI assistant and an MCP server exposing 18+ tools for external AI integration. Works with any OpenAI-compatible LLM — no vendor lock-in.
About This Project
AI Charts combines a conversational AI assistant for rapid chart creation with a composable MCP server that lets external AI tools like Claude Desktop and Cursor create and manage diagrams programmatically. Supports flowcharts, ERDs, swimlane diagrams, mind maps, and sequence diagrams with ELK-based auto-layout, structural validation, cycle detection, and export to Mermaid, Markdown, and PDF.
Key Achievements
- Designed and built a composable MCP server with 18+ tools enabling any AI assistant to create and manipulate diagrams programmatically
- Implemented agentic tool-calling loops where the AI assistant iteratively builds charts through multi-step function calls — add nodes, edges, groups, validate, and auto-layout in a single conversation turn
- Built with full LLM provider abstraction — works identically with Ollama (local), OpenAI, Groq, or any OpenAI-compatible endpoint with zero code changes
- ELK-based auto-layout engine with cycle detection, reachability analysis, and structural validation for complex diagram types
- Full revision history with event-sourced undo/redo tracking across all operations
- Portable ZIP project archives and multi-format export (Mermaid syntax, Markdown summaries, PDF)
Related Projects
Related Publications
Articles that cover technologies and skills used in this project

Prop Drilling is a Code Smell When Used Incorrectly
Prop drilling is considered an anti-pattern in React because it involves passing props down multiple levels of a component tree, even when only a few components need them. This can lead to several...

RPC; Everything Old is New Again
Possibly since the inception of computing, the paradigm of client to server communication has been an important, sometimes overlooked requirement for distributed computing. From a developer...
Benchmarking Models in the Real World
I tested a bunch of SOTA LLM models on real world agentic tasks that go beyond coding and typical business processes and really stress them. There were some shocking...