Free AI Safety Workshop — in collaboration with explainx.aiSave your seat
Agentbeam
Beam Sentinel

The engine behind Agentbeam.

Beam Sentinel is the local-first security engine that records AI agent activity, flags risky actions with evidence attached, and scans MCP configs and SKILL.md files against heuristic patterns — before anything reaches the model or leaves your machine. It runs two ways: as the runtime engine that powers Agentbeam, and as a free Chrome extension for AI chat assistants.

11

heuristic patterns per MCP/SKILL.md scan, plus a version-pin check

6

risk categories checked on every AI chat send, browser-side

0

network requests required for either product to work

Two surfaces, one engine

Same idea, wherever the risk shows up

Agent activity and AI chat prompts fail the same way: something risky slips out before anyone reviews it. Sentinel catches that moment on both surfaces.

Timeline of observed AI agent activity recorded by Beam Sentinel

The runtime engine

Sentinel for your AI agents

The engine that powers Agentbeam. It records shell commands, file edits, tool calls, and network activity from your coding agents into a local timeline, flags risky actions with the evidence behind them, and scans MCP configs and SKILL.md files before you trust them.

  • Runtime activity recorded as local NDJSON files — no cloud service, no telemetry
  • 11 heuristic patterns plus a version-pin check for tool poisoning, rug pulls, and credential exfiltration
  • Works across Claude Code and other terminal-based or IDE agents
See the full platform
Beam Sentinel Chrome extension coverage grid listing six risk categories

The browser guardrail

Sentinel for your browser

A free Chrome extension that checks what you're about to paste into an AI chat assistant before it leaves your browser — a live secret gets redacted automatically, other high-risk sends pause for review.

  • Free, no account, no paid tier
  • Checks run entirely on your device — no network request, ever
  • Only the last 100 redacted decisions are kept, locally
Get the Chrome extension

FAQ

Questions about Beam Sentinel

What is Beam Sentinel?

Beam Sentinel is the local-first security engine behind Agentbeam. It records AI agent activity, flags risky actions with evidence attached, and scans MCP configs and SKILL.md files against heuristic patterns — all on your own machine, with no cloud service and no telemetry by default.

Is Beam Sentinel one product or two?

Two surfaces, one engine. The runtime engine watches agent activity on your machine and powers Agentbeam directly. The Beam Sentinel Chrome extension is a separate, free browser tool that applies the same idea — catch risk before it happens — to what you paste into an AI chat assistant.

Is Beam Sentinel free?

The Chrome extension is free, with no account and no paid tier. The runtime engine ships as part of Agentbeam, which is local-first and free to run today, with a licensed self-hosted control plane available for teams.

Does Beam Sentinel send data anywhere?

No, by default. Both surfaces run their checks locally — the runtime engine writes to local NDJSON files, and the Chrome extension makes no network request at all.

Start with the free extension

The fastest way to see Sentinel at work — no account, nothing to configure.

Get the Chrome extension