[garethx]

Requested by Gareth · project thread

Before: hookdeck/webhook-registry#105 (verified sources batch 3, merged) added 44 provider rows, upgraded Smartsheet's, and exported doc examples for 30 of those 45 to the registry's samples-doc/ tree. None of the 30 had a directory here.

After: 30 new provider directories with 228 doc-sourced samples. providers.json goes from 210 to 240 providers. The other 15 rows have no published example body, so they get no directory: breezy-hr, checkly, duffel, exa, expo, fourthwall, gitea, heygen, lago, luma, mautic, propelauth, sendcloud, smartsheet and vgs. No existing file is changed. This touches only new directories.

Provider	Samples	Version dir	topic_identifier
ably	3	latest	items[].source
acuity-scheduling	1	latest	action
bird	19	latest	type
bland-ai	1	latest	event_type
boldsign	7	latest	event.eventType
census	3	latest	event
check	16	latest	null (composed: Check-Topic + . + event)
cloudtalk	22	v1	type
codat	23	latest	eventType
descope	2	latest	[].action
doppler	1	latest	type
firecrawl	14	v2	type
firehydrant	2	latest	event.resource_type
flagsmith	1	latest	event_type
go1	2	3.0.0	event_type
granola	1	latest	event_type
help-scout	1	latest	x-helpscout-event (header)
jfrog	24	latest	null (composed: domain + . + event_type)
kit	1	latest	events[].type
kombo	7	latest	type
kustomer	10	latest	name
loop-returns	3	latest	trigger
mailerlite	14	latest	list: event, type, events[].type
seatable	3	latest	data.op_type
stedi	12	v1	type
taiga	14	latest	null (composed: type + . + action)
terraform	4	latest	list: notifications[].trigger, trigger
toggl	1	v1	null (no discriminator)
waha	11	latest	event
yapily	5	latest	type
Total	228

How: I copied samples-doc/providers/<p>/ from webhook-registry main (5a396b63) into providers/, the same way #70 did. Every sample file is byte-identical to the export. Two index.json files differ, and those changes are listed first below. I checked each topic_identifier with a port of resolveTopic against every sample, and every non-null one resolves all of its provider's samples. Each file carries source.type: vendor-documentation and its page URL, and each version is marked sourced_via: docs. yarn compile passes (210 → 240 providers, compiled topic count equals file count for all 30). yarn test passes 16/16.

Worth a look:

• mailerlite uses the README's list form. The export wrote event, which resolved 7 of 14 samples. The registry row records three carrier families (event, type, batched events[].type), and the automation, group, campaign-click/open and delete bodies carry type instead of event. Listing them in registry order resolves all 14 (unipile and veriff precedent).
• terraform uses the list form too. The registry's primary carrier is notifications[].trigger (run notifications, payload_version 1), with top-level trigger for workspace and assessment notifications (payload_version 2). The export keyed on the second family's trigger alone, which resolved 2 of 4 (run:errored and verification were missed). The list resolves all 4.
• Composed rows are null: check, jfrog and taiga, as exported. This follows the modern-treasury, stytch and webex precedent from Add 186 doc-sourced samples for verified-sources batch 2 (26 providers) #70. toggl is null because the registry records no discriminator (its one sample is the ping).
• Versions taken from the payload: cloudtalk v1 (body version), stedi v1 ("object": "v1.event", the eventPayload.v1Event schema) and go1 3.0.0 (body webhook_version). These follow the README rule.
• Versions kept as exported but open to question:
  • firecrawl v2: there's no version field in the payload or headers. The registry files it under v2 only because the webhook object is documented on the /v2/ endpoints.
  • toggl v1: the registry says versioned: false. The v1 is the Webhooks API path (/webhooks/api/v1), and the docs URL carries no version. In Add 186 doc-sourced samples for verified-sources batch 2 (26 providers) #70, ramp stayed latest for the same reason, so going by that, toggl should be latest. It's a one-line change plus a directory move.
  • ably latest: every delivery sends an x-ably-version: 1.2 header, which the docs call "the version of the Webhook". The README rule names payload fields, so I left it at latest. Following the castle precedent, it would be 1.2.
  • terraform latest: its bodies carry payload_version 1 and 2. These mark two notification shapes that are both current, so they aren't successive versions. Splitting them would hide one shape behind a non-latest directory.
• bland-ai keys on event_type, not the registry's primary category. The registry records category, then event_type. The one sample (citations) carries only event_type, so category alone would resolve 0 of 1. It stays as exported, like didit in Add 186 doc-sourced samples for verified-sources batch 2 (26 providers) #70, where every sample resolves. ["category", "event_type"] would match the row and is a one-line change. ably and codat also have unlisted extra families, but all of their samples resolve, so they stay strings.
• Filenames differ from topics in three providers. descope's topic is Change Password (spaces on the wire), kombo's is assessment:order-received, and terraform's triggers use : (run:errored, assessment:drifted, workspace:auto_destroy_reminder). The files are Change.Password.json, assessment.order-received.json, run.errored.json and so on, but the published topic keys come from each file's topic field and match the wire values.
• descope [].action is a root-level array. It resolves now that the resolver fix from Add 147 doc-sourced samples for the verified-sources pilot (10 providers) #68 is merged.
• census cites fivetran.com docs, because Census is now part of Fivetran.

---

Generated by Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
webhook-samples	Ready	Preview	Oct 22, 2026 11:05am UTC