Skip to content

Platform, security, and engineering leadership

You cannot govern what you cannot enumerate.

Agents arrive through IDE extensions, side projects, and vendor tools. Most organisations discover their true agent inventory during an incident. Cirvix builds it from observed traffic instead.

Ask yourself

How many MCP servers are configured across your engineering laptops right now?

Today

What you are working around

  • No canonical list of agents, and any list you build is stale within a week
  • Agents running under a developer's personal credentials
  • MCP servers added to individual machines, each a separate trust boundary
  • Shadow spend on model APIs that no one can attribute

What changes

The specific difference

Each of these is a module you can read about in detail — this is the same platform, not a variant.
  • Discovery from traffic

    Agents appear in the inventory because they made a call, not because someone remembered to register them.

    Analytics
  • Ownership as a work queue

    Agents without an accountable owner are surfaced as a list to resolve, rather than quietly assumed to be fine.

    Agent Identity
  • A registry for MCP servers

    Every server an agent can reach is registered, pinned, and inspected — including the ones already on developer machines.

    MCP Gateway
  • Attribution for spend

    Model and tool consumption attributed per agent, per team, per environment.

    Analytics

In detail

Start with the scanner

`npx @cirvix/agent-control scan` is read-only, needs no account, and runs on one machine. It reports the agent runtimes, MCP server configurations, and credential files reachable from agent context. Running it on a handful of engineers' laptops is usually enough to change the conversation about how large the problem is.

The inventory has a known gap, and it is reported

Cirvix sees agents whose calls traverse it. Agents configured to bypass the gateway entirely are, by definition, not in the traffic — which is exactly why the scanner exists as a separate, host-level tool. Coverage is shown per host rather than as one reassuring number.

Get started

Let the agents do real work.

Start with the free local scanner — it needs no account and tells you what is already ungoverned on your machine. Move to a governed runtime when you are ready.

Cirvix AgentControl is in private beta. Pricing is published and plans are not yet chargeable — early access confirms your place and we will tell you before anything bills.