Beacon
Endpoint inventory & governance
A signed, user-space agent that inventories and governs where agents run.
A signed, user-space daemon on every machine, with no kernel driver and nothing to break. It inventories the AI layer (agents, CLIs, versions, MCP servers), flags hygiene and exposure, and evaluates policy on every MCP call locally. Monitor first, enforce when you're ready.
- Inventories installed AI tools, versions, and the MCP servers they're wired to.
- Flags hygiene and exposure: down-level tools, unpinned MCP servers, secrets in config.
- Evaluates policy on every MCP call at the endpoint, where the call originates.
You can't govern the AI agents you can't see.
Engineering funded a Claude Code, Cursor, and Codex rollout, and now every laptop runs AI agents wired to MCP servers nobody inventoried. Your EDR sees processes, your scanner sees packages, your IdP sees logins. None of them can tell you which agents are running, which tools they can invoke, or what they did. That blind spot is where your next incident starts.
The questions your reviewers will ask.
“I don’t know what’s installed.”
Beacon inventories every AI agent, CLI, and version on each endpoint, and the MCP servers each one trusts, across every vendor.
“I can’t see the risky configurations.”
It flags hygiene and exposure: down-level tools with known CVEs, unpinned MCP servers, and secrets left in config.
“I have no control over what an agent can do.”
Beacon evaluates policy on every MCP call locally, at the endpoint where the call originates, with no cloud round-trip.
“I can’t put another kernel agent on these machines.”
It runs as a signed, user-space daemon. No kernel driver, no hooking, no conflict with your EDR.
“I can’t risk breaking the developer rollout.”
It deploys read-only and starts in monitor mode, so you see what enforcement would do before anything is blocked.
Evidence, not assurances.
What a security leader walks away able to demonstrate to a board or an auditor.
- A board-ready inventory of every AI agent, tool, version, and MCP connection across the fleet, in about 30 minutes.
- A ranked view of exposure: which endpoints run vulnerable or unsanctioned AI tooling.
- A record of what enforcement would have done, before you turn it on.
- One command removes it. Nothing to rip out.
- Platforms
- macOS, Linux, Windows
- Footprint
- Signed user-space daemon, no kernel driver
- Deploy
- Read-only, about 30 minutes
- Enforcement
- Local, monitor-first, opt-in
- Removal
- One command
One module of the Alyria platform.
Beacon works alongside the rest of Alyria — prevention and detection for every AI agent your people run, tied together by one policy engine and one signed audit chain.
Put Beacon to work.
Deploy Beacon read-only and see how Beacon fits at the endpoint, under your keys.