LandfallDocs
Triggering integrations · Vendor Status Watch

Opening a war room from a vendor's own status

Unlike every other page in this section, there is no webhook payload to configure on a third party's side. Vendor Status Watch is a background poll inside Landfall — the "signal" originates from Landfall reading a public status feed, not from an external caller reaching in.

Cloudflare/Datadog/GitHub/Stripe
publish a disruption on their own status page
Landfall's own poll
every 5 minutes, matched against your Architecture Map
Landfall
opens or attaches an incident

How it works

Set a provider's response level to Open incident automatically on the Vendor Status Watch settings page — that is the whole configuration. There is no route to POST to, no secret to issue, no payload shape to match. Once configured, the very next poll cycle that finds a genuine match opens (or attaches to) an incident automatically, exactly like every other automated source's trigger — reusing the same underlying open/attach/repeat/recovery mechanism (see Integrations architecture) — it just never crosses an HTTP boundary to get there.

A response level of Notify only (the default) never opens automatically; a human's explicit Declare incident click on the resulting notification is what calls the same mechanism, on demand, instead.

Deduplication and recovery

A provider's own advisory id is the attach identity: repeated polls of the same still-open disruption attach to the one incident rather than opening another, and a provider's own "resolved" update is recorded on the incident (never silently swallowed) without closing it — recovery of your own systems stays a human decision, the same rule every paging-tool integration in this product already follows.

Reference

FieldValue
Trigger originInternal (Landfall's own poller) — not a webhook, no inbound route
ConfigurationResponse level auto-incident on the watch settings page, per provider
Attach identityThe provider's own advisory id
Replay identityThe advisory id plus the provider's own last-updated timestamp — an unchanged re-poll is a true no-op
RecoveryRecorded on the incident; never auto-resolves it