Mehdi Belckadi

Python, static analysis tooling, OpenAPI/JSON Schema, FastAPI. Some LLM tooling experience, kept deliberately out of the core scanner. Generalist backend engineer, security-focused right now.

How MCP servers fail in production once agents have real write access, and how to explain agent-specific risk to founders without sounding like FUD.

Nexum — a deterministic scanner flagging what could go wrong if an AI agent gets real tool-calling access to your API. No LLM in the core detection path.

Talking MCP/agent security, OpenAPI spec design, or comparing notes with other solo technical founders. Run an MCP server? Happy to swap scan findings.