Twitter Webhook API: Real-Time Post Alerts, Free to Call

Two free endpoints. POST /twitter/webhook registers an HTTPS URL you control and returns an HMAC signing secret, shown once, that you use to verify deliveries. POST /twitter/monitor then watches an X account and, once you point it at a webhook, POSTs a signed tweet.created event to that URL every time the watched handle publishes. There is no backfill: only posts made after the monitor was created are delivered. Both endpoints are zero-rated, they never consume API credits.

Two separate numbers matter. Detection is bound by the shared poll interval, currently around 60 seconds across every plan and not a published SLA, so a new post can take up to about a minute to be seen. Delivery, the time from us detecting the post to our POST landing on your endpoint, is typically a couple of seconds on the Growth plan and above; on a real production event measured after shipping this feature, detection took about 34 seconds and delivery took about 3.6 seconds, for roughly 38 seconds end to end. This is a fast-polling architecture, not a firehose, and it is not marketed as sub-second push.

Every delivery is signed HMAC-SHA256 over the timestamp and the raw request body, carried in an X-TwitterAPIs-Signature header. Recompute the signature on your end using the secret returned once at webhook creation and compare it to the header before trusting the payload. A one-shot test delivery (POST /twitter/webhook/{id}/test) sends a real signed event with event set to webhook.test rather than tweet.created, so you can confirm your receiver and signature check both work before relying on it.

Creating or resuming a monitor can return 503 capacity_unavailable if the shared pool cannot currently serve one more monitor at the published interval. Retry later; this is a capacity signal, not a rejection of your account or your webhook.

No. add_user_to_monitor_tweet, remove_user_to_monitor_tweet and get_user_to_monitor_tweet are compat drop-ins that translate an x_user_stream-shaped request onto the same underlying monitor system, so a migration can point at the compat routes first and move to monitor/webhook directly on your own schedule. Both paths bill the same, which is to say nothing.

Every call to create, list, update or delete a webhook or a monitor is $0, including the health and delivery-log endpoints. What is not free is watching more than one account at a time: the first monitor on any account is a free slot, and watching more requires a monitoring plan, Starter at $15/mo for 6 slots up to Scale at $799/mo for 2,000, billed as a flat monthly subscription rather than the per-call model the rest of the API uses. See the full breakdown on the pricing page's monitoring section.

No. You need only a TwitterAPIs Bearer token, issued after a 30-second signup. There is no X developer application, no app review, and no enterprise sales process, which is the gate that keeps X's own first-party Account Activity webhooks out of reach for most builders.

Not yet through the webhook path. A monitor watches one X handle, optionally narrowed with a domain_filter (deliver only posts linking to a given host) or include_replies (turn reply delivery off). For keyword or hashtag tracking, run GET /twitter/search/tweet-advanced-search on a polling loop instead, using the same Bearer key at $0.0008 per call, see the Twitter search API.

Three read endpoints, all free. GET /twitter/monitor/{id}/health returns one monitor's cursor: its last seen tweet id, last poll time, and a degraded flag. GET /twitter/monitor/health is an account-wide rollup with no id needed, active and paused counts plus pending, delivered and failed deliveries over the last 24 hours. GET /twitter/monitor/deliveries lists your most recent delivery events with real measured detected_lag_ms and delivery_lag_ms per event, so you can audit actual latency rather than trust a published number.