Dan Arpino is a software engineer at Snyk developing agentic development security for autonomous coding assistants. He helped build the company’s Evo Agentic Development Security platform and created Snappy, an experimental desktop application that monitors coding agents and enforces security policies directly on developers’ machines.
Arpino graduated from the University of Massachusetts Amherst in 2007, completing a thesis on coding theory and the minimum distance of cyclic codes. He subsequently worked at Vista and, at Snyk, built an agentic-development team. His professional profile reflects his emphasis on securing AI-generated code as it is written, before vulnerabilities propagate through development workflows.
After working on integrations that scanned agent-generated code, Arpino built Snappy as a local Electron application to observe the broader development environment. The prototype inventories running models, Model Context Protocol servers, skills, and command-line tools; records agent sessions, file access, executed commands, and vulnerabilities; and can launch coding agents to repair detected problems.
His approach centers on three practical principles:
- Local visibility into agent behavior: Developers should know which agents and tools are active, what files they access, and which commands they execute.
- Deterministic security guardrails: Enforcement should block unauthorized actions directly, including attempts to read sensitive local files, without depending on a model’s judgment.
- Workspace-specific security policies: Controls should adapt to individual projects; when automatic remediation repaired deliberately vulnerable tools he needed for benchmarking, Arpino disabled it for that workspace.
At AI Engineer World’s Fair 2026, Arpino demonstrated Snappy alongside Snyk product director Ezra Tanzer, presenting it as an experimental prototype rather than a committed commercial release.
