Your agents get tools, not tabs.
Both Virex products expose a first-party Model Context Protocol server. Claude Code and other MCP-compatible clients pick the tools up and use them the way they use any other tool, without a human switching windows.
// Built by Virex. Not a community bridge.
One protocol, both products.
MCP is the standard way an agent discovers and calls tools. Virex ships a server for each product, so the agent that writes your code is the same agent that reads your memory and reports its own session activity.
Virex Memory over MCP
Agents write memories as they work and search them later by meaning. Retrieval is ranked by a relevance score that is refreshed on a schedule, so what surfaces reflects what still matters.
Virex Remote over MCP
Session activity, permission requests, and plan decisions are exposed as structured records, so an orchestrating agent can follow a run and act on what it finds.
Memory follows the organization, not the machine.
An MCP tool call is portable. That is the whole point of putting these products behind one.
Context survives the session
What an agent learned on Monday is searchable on Thursday, from a different machine, by a different teammate, inside the same organization.
The agent asks, rather than you pasting
When an agent needs prior context it searches for it directly. Nobody has to maintain a briefing document that goes stale the week it is written.
Access is scoped by seat
MCP access respects the same entitlements as the web portal. A member without a seat on a product does not get that product tools.
Where these tools show up.
Claude Code
The primary integration for both products. Virex Remote is designed around Claude Code sessions specifically, and Memory is available to Claude Code as a standard MCP tool set.
MCP-compatible tooling
Virex Memory works with other agent tooling that speaks MCP. The tools are the same, and so is the organization-scoped memory behind them.
Your own agents
If you build agents yourself, Virex Memory also offers first-party SDKs and REST and gRPC APIs on one host, with the same tenancy and access rules.
What the MCP servers do not do.
The security model does not loosen because a tool call is involved.
They do not execute your code
The servers coordinate and carry state. Execution stays on your machines, under your own model subscription.
They do not cross tenants
Every tool call resolves inside the calling member organization, enforced at the database layer with row-level security.
They do not bypass seats
Tools appear for members holding a seat on that product. Removing a seat removes the access with it.
MCP questions
What is the Model Context Protocol?
MCP is an open protocol that lets an AI agent discover and call external tools in a standard way. Instead of an integration written per client, a tool server describes what it offers and any compatible agent can use it.
Which Virex products expose an MCP server?
Both. Virex Memory exposes memory reading and writing. Virex Remote exposes session activity and the decision surface for sessions running on your machines.
Do I have to use MCP to use Virex?
No. Both products have full web interfaces in the product portal. MCP is how your agents use them without you in the loop.
Does an agent using MCP see other teams data?
No. Tool calls resolve within the calling member organization only, and isolation is enforced at the database layer rather than in application logic alone.
Where do I find setup instructions?
Configuration and reference material live in the docs. This page covers what the surface does rather than how to wire it up.
Does Virex run the agent for me?
No. Agents run on your machines. Virex coordinates the session, stores what it is told to store, and relays your decisions.
Give your agents the tools.
Subscribe, assign a seat, and the MCP surface for that product is available to the member holding it.