An open-source scanner that inspects packages from 28 ecosystems (PyPI, npm and others), AI-agent skills and MCP servers for malware before they install. It runs locally with no account and no telemetry. Every block shows its evidence (file, behaviour, rule), and it fails closed when an attacker tries to hang or crash the scan. Detection combines more than 1,500 open rules, cross-file data-flow analysis and small local classifiers. Funding pays for continuous measurement on held-out malware, faster capture of new attacks, and reporting new malware to OpenSSF and the registries.