threatdeck
Live 13:15 UTC ·

TrustLast updated 2026-11-10

Security

ThreatDeck is built to hold as little as possible. There are no accounts or passwords; the data we do keep is listed on the privacy page. This page explains how it is protected and how to report a problem.

Hosting

  • The site, REST API and MCP server run on Cloudflare (static assets and a Worker). Alert channels, fleets, API keys and business enquiries live in Cloudflare D1; backups in Cloudflare R2. Cloudflare encrypts this storage at rest.
  • The news pipeline and newsletter run on the operator's own server in Docker, with no inbound ports: admin interfaces are reached through Cloudflare Tunnel behind Cloudflare Access.
  • The pipeline only reads public sources. It holds no visitor data except newsletter email addresses.

Encryption in transit

Pages, API and MCP server are served over HTTPS only (TLS by Cloudflare); the content security policy upgrades any insecure request. Alert deliveries go to Slack, Microsoft and Discord over HTTPS.

Tokens and keys

  • Alert manage tokens (192 bits), fleet admin and upload tokens (192 bits) and API keys (256 bits) come from a cryptographic random generator, are shown once, and are stored only as SHA-256 hashes. We cannot see or recover them.
  • Fleet upload tokens can only add inventory; the admin token reads and deletes. The upload token can be rotated.
  • An unknown, malformed or revoked API key is rejected (401), never silently treated as anonymous.
  • ThreatDeck never stores IP addresses, only a shortened SHA-256 hash used for abuse limits.

Minimal data, enforced by the server

  • Scanner and fleet. threatdeck scan runs locally and uploads only with an explicit --upload token; --offline forbids network use. The upload is an allowlist: a random machine id (stored hashed per fleet), an optional label, OS family, CLI version, tool slugs and versions, MCP client, transport, package id and public host, and finding counts. The server schema rejects any other field with an error and stores nothing; paths, usernames, hostnames, config contents, environment values and transcripts are never accepted. Details on /fleet/.
  • Alerts. Webhooks are accepted only for Slack, Discord and Microsoft Teams Workflows hosts with fixed path shapes, a test message must succeed first, redirects are never followed, and messages carry titles, one-liners and ThreatDeck links only. The full webhook URL is never returned by the API.
  • Abuse limits. 60 API requests per minute per address (600 with an API key), and at most 5 create attempts per address per hour for alert channels, fleets, API keys and business enquiries.

Web application

  • A strict content security policy (no plugins, no framing, scripts only from this site plus Cloudflare's Turnstile and optional Web Analytics), X-Content-Type-Options: nosniff, X-Frame-Options: DENY, a referrer policy and a permissions policy that turns off camera, microphone, location and payment APIs.
  • Source content is treated as data, never as instructions, and escaped when rendered. Dashboards render API data as text, not HTML.
  • No third-party trackers or ad scripts.

What we do not have yet

ThreatDeck has no SOC 2 or ISO 27001 certification, no single sign-on, roles or audit log (access is by token), and no uptime commitment. The scanner is one unminified file served from this site with a published SHA-256 checksum (/scan/); it is not on npm yet, and an npm release will come with provenance. If your review needs something not on this page, ask: hello@threatdeck.net.

Reporting a vulnerability

Email hello@threatdeck.net with what you found, the affected URL or component, steps to reproduce and the impact. Machine-readable contact: /.well-known/security.txt.

  • In scope: threatdeck.net, the REST API (/api/v1/), the MCP server (/mcp), team alerts, the fleet view, and the threatdeck scan CLI.
  • Please do not: access or change data that is not yours, run denial-of-service or volume tests, send spam through alerts or forms, or use social engineering. Use your own alert channels, fleets and keys for testing.
  • What we do: acknowledge your report within 3 business days, keep you updated, fix confirmed issues as fast as we can, and credit you if you wish. We will not pursue good-faith research that follows this policy. There is no paid bounty.

If something goes wrong

If an incident affects data we hold, we will post a notice on this page and email the people we have an address for, without undue delay, saying what happened, what data was involved and what we are doing about it. Alert channels and fleets have no email on file: if you want to hear from us, send your address through the business contact form or by email.

Related: Privacy · Terms · Methodology

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.