threatdeck
Live 13:15 UTC ·

Fleet view

Which AI coding tools and MCP servers run on your developers' machines, which versions are below the minimum safe version, and which MCP packages have advisories. Each machine runs threatdeck scan --upload; only inventory is sent, never paths, names or config contents.

Free, no billing, up to 500 machines per fleet. Machines that have not uploaded for 30 days drop off.

Create a fleet

Dashboard

Exactly what a machine uploads

Only with --upload <token> or THREATDECK_FLEET_TOKEN; --offline forbids it. The CLI prints one line saying what it sent, and the server rejects any upload with a field outside this list.

  • A random machine id, created once at ~/.config/threatdeck/machine-id (not derived from the hostname, MAC address or username; stored on the server only as a hash).
  • Your --label (optional, up to 60 characters), the CLI version and the OS family (macOS, Linux or Windows).
  • AI coding tool slugs and versions, e.g. claude-code 2.1.319.
  • For each enabled MCP server: the client, the transport, the package id (npm:@scope/server@1.2.3) when the scanner would already send it to the lookup API, and the remote host only when it is public.
  • Finding counts by rule id and severity (for example 2 × TD-MCP-UNPINNED low), never finding text.

Never sent: file paths, usernames, hostnames, server names, commands, arguments, env or header names and values, config contents, transcripts.

Following

Follow the vendors, products, AI tools, techniques and actors you care about. Matching stories, CVEs and exploited bugs are marked, a Following tab appears in the feed, and returning visits show what changed. Saved only in this browser.

Presets

Get updates

A personal RSS feed of new stories on everything you follow. Paste it into any feed reader, Slack or Teams RSS app.