AI Development

Nyro MCP

Secure MCP server that lets AI agents work with real files and shell workflows without breaking safety boundaries.

PythonMCPAIFilesystem

Challenge

AI agents are only useful when they can touch real systems, but that often introduces serious risk.

I designed strict guardrails first: filesystem sandboxing, controlled shell execution, explicit boundaries, and full traceability.

Nyro MCP proves agent workflows can be practical and safe in real development environments.

Includes root path isolation, fast glob search, ZIP handling, command timeouts, and detailed runtime logs.

Neuromorphic research platform for running large spiking neural network simulations on GPU in real time.

Prompt framework that turns generic assistants into specialized agents for real professional work.