Discover shadow MCP
Atlas finds MCP connections in your code, even before anyone registers the server.
Atlas finds MCP connections in your code, even before anyone registers the server.
AI SECURITY
MCP Gateway
Varonis Atlas governs the MCP servers and tools your agents use. Find shadow MCP, approve the tools you trust, and block the ones you don't.
Trusted by 8,000+ customers
OVERVIEW
MCP gives your agents tools and data.You decide what they get.
Model Context Protocol (MCP) servers give your agents new actions and new access to your data. Atlas finds every server, shows what each one can reach, and enforces your allowlist at runtime.
Control tool calls
Build allowlists of reviewed tools. Atlas strips everything else before the model sees it.
Build allowlists of reviewed tools. Atlas strips everything else before the model sees it.
Require explicit credentials
Bind a credential to each server on a gateway token. No binding, no access.
Bind a credential to each server on a gateway token. No binding, no access.
Reduce your blast radius
Focus on the permissions and exposed data that could increase the impact of a breach.
Accelerate compliance
Bring data discovery, access visibility, and activity monitoring together to support your compliance program.
DISCOVERY
See every MCP server and the data it can reach
Atlas finds MCP connections in your code and inspects each server to list its tools. See which servers can reach sensitive data, who owns them, and how risky they are. Shadow MCP connections show up for you to register or link.
Varonis product overview
auto
image
Allowlists
Allow only the tools you approve
Create a Virtual MCP for each workflow and switch on the tools it needs. Draft changes, then publish. Decide whether new tools from a server get added on their own or wait for review.
Varonis product overview
auto
image
RUNTIME
Hide unapproved tools from your agents
MCP Quarantine compares the tools in each request to your allowlist. It removes any tool that isn't on it, so the model never sees it.
Varonis product overview
auto
image
Monitoring
Catch tool drift and risky changes
Policies flag shadow MCP, changed tool definitions, new tools on approved servers, and servers that reach sensitive data with no owner. Activity shows which tools get called and which get quarantined.
Varonis product overview
auto
image
FAQs
Common questions
Does registering a server allow its tools?
Does registering a server allow its tools?
No. Registering gives you visibility. Tools are allowed only through a published Virtual MCP.
Which MCP transports are supported?
Which MCP transports are supported?
Streamable HTTP, Server-Sent Events, and STDIO.
What if two Virtual MCPs apply to the same resource?
What if two Virtual MCPs apply to the same resource?
Their allowlists add up. A tool is allowed if it's approved and in any Virtual MCP that applies.
Which credentials can I use?
Which credentials can I use?
API keys or OAuth. OAuth credentials refresh on their own. Rotate an API key and your token bindings stay intact.
Can a token reach a rejected server?
Can a token reach a rejected server?
No. Rejected servers are blocked and can't be bound to a token.
Can I set MCP policies once for the whole company?
Can I set MCP policies once for the whole company?
Yes. Turn on a policy at the top level and it applies to every organization and project below. Teams can change it lower down.
Does this cover LLM calls too?
Does this cover LLM calls too?
The LLM Gateway applies the same governance to your LLM traffic.
Ready to see Varonis in action?
https://cdn.prod.website-files.com/6aad3cd63e81c645c771a0be/6aad48e61926ad27c8821c7b_varonis-conversion-blue.mp4
https://www.youtube-nocookie.com/embed/717uPD8k5nc