mastyf.ai
GitHubDownload ShieldDownload

One checkpoint between your agents and your tools

Mastyf is five parts built around that checkpoint. Select a part to read how it works.

Control PlaneOne policy for every agent, with signed audit trails.
Your agentsClaude, Cursor, LangChain, or your own
Mastyf checkpointShieldOn desktops: guards Claude Desktop, Cursor and other local agents.GatewayIn production: a fail-closed proxy in front of your agents.Blocked calls stop here. Zero bytes reach a tool.
Your toolsFiles, shell, network, databases
SwarmAttacks your policy in CI, before an attacker does.TrustScores the MCP servers you install.
Agents send tool calls to the Mastyf checkpoint, made of Shield on desktops and Gateway in production. Only allowed calls reach your tools. Swarm tests the checkpoint’s policy in CI, Trust scores the tool servers, and the Control Plane governs every agent.
Download ShieldRead the component docsHow Mastyf compares