Security

Security and abuse prevention

Webhooks carry credentials and personal data, and a public tool that accepts and sends HTTP requests attracts misuse. This page explains what the service does about both.

Protecting what you capture

Webhook URLs are write-only

The URL you give a provider can only receive. Reading requests or changing settings needs a separate dashboard key. The key is derived from the URL's ID with a server-side secret, lives in the fragment of your dashboard link and in your browser's storage, and is sent only in an Authorization header. A URL that leaks through a provider's logs or a screenshot does not expose your data.

IDs cannot be guessed or forged

Each ID contains 80 random bits, its own expiry time and a cryptographic signature. The edge checks the signature before doing anything else, so requests to made-up IDs are rejected immediately and never reach storage.

Secrets used for verification stay in your browser

Signature checks run in the page with the Web Crypto API. Signing secrets are not transmitted, logged or stored by the service.

No third-party code where your data is shown

The inspector loads scripts from this site only. Its Content-Security-Policy forbids everything else: no analytics, no advertising code and no external fonts. Banner ads appear on content pages only.

Deletion is automatic

Every URL has an expiry. When it passes, the captured requests, settings and delivery logs for that URL are erased. You can delete a URL sooner, which takes effect at once.

Isolation and transport

Each URL's data is held in its own isolated storage instance on Cloudflare's network, encrypted at rest. All traffic uses HTTPS.

Preventing abuse

Rate limits at the edge

  • Creating URLs: 6 a minute and 30 a day per client, with an optional human-verification challenge.
  • Receiving: 600 requests a minute per URL, and 25,000 over a free URL's lifetime.
  • Probing: clients that repeatedly request invalid IDs are throttled.
  • Size: bodies are read up to a fixed ceiling and never buffered without bound. A URL keeps a bounded number of requests and bytes.

Responses cannot become web pages

Custom responses cannot use an HTML content type, set cookies or redirect. They are served from a separate origin with a policy that tells browsers to treat the content as inert. A test URL cannot be turned into a phishing page or a link shortener.

Outbound deliveries are fenced in

Automations make requests on a user's behalf, so they are restricted:

  • Destinations must be public HTTPS hostnames on port 443. IP addresses, private and local names, embedded credentials and this service itself are refused, when saving and again when sending.
  • Redirects are not followed.
  • Each URL has per-minute and per-day delivery budgets. Deliveries are never retried, so the service cannot amplify traffic.
  • An automation that keeps failing pauses itself.
  • Requests that originate from an automation never trigger further automations.
  • Every delivery carries a User-Agent and X-Webhook-Stream-Endpoint header naming its source.

Reporting abuse or a vulnerability

If a webhook.stream URL is being used against you, or you are receiving unwanted deliveries, write to abuse@webhook.stream with the URL or the X-Webhook-Stream-Endpoint header value. Offending endpoints are disabled and their data erased.

To report a security vulnerability, use the same address. Please give us a reasonable chance to fix the problem before disclosing it, and do not access data that is not yours while testing.

Using the service sensibly

  • Prefer providers' test modes and test secrets.
  • Avoid sending real customer data or live credentials to any testing tool, this one included.
  • Share dashboard links only with people who should see the data.
  • Delete URLs when you have finished with them.