Skip to main content
The ingest pipeline is Citable’s backend path for high-signal edge traffic: classify at the edge → POST to ingest → queue → ClickHouse → Agent Analytics. There are two ways to get data into this pipeline:
Managed partners contact Citable before setup — we operate the edge stack. Self-hosted teams self-provision ingest credentials in Connectors; contact us only if you want hands-on help.

​
What the pipeline does

Every classified request produces telemetry in one or both stores: Analytics Engine powers live counts (overview, time series, bot breakdown). ClickHouse holds row-level events for lag analysis, joins, and long retention.

​
Architecture

The edge layer always forwards requests to your origin. Telemetry is fire-and-forget and does not block responses.

​
Authentication

Ingest accepts POST https://ingest.getcitable.com with header x-auth-token. We store token:{token} → siteId in Cloudflare KV. Ingest returns 401 for missing, invalid, or revoked tokens.

​
Event payload

Minimum fields your Worker (or our edge proxy) must send:
siteId must match the value Citable registered when we provisioned your credentials.

​
Choose your path

​
Managed — trusted partners

Best when you want Citable to operate the full stack:
  1. Contact Citable to request managed proxy access
  2. Follow AI Traffic Proxy (Managed)
  3. Add CNAME → proxy.getcitable.com, verify in Connectors
No Worker deployment, no ingest configuration on your side.

​
Self-hosted — your Cloudflare zone

Best when you already manage Cloudflare Workers:
  1. Open Connectors → AI Traffic Proxy → Self-hosted Cloudflare Worker
  2. Copy siteId and ingest token into your Worker (see Cloudflare Worker guide)
  3. Deploy on your zone
You operate the Worker; Citable operates ingest → ClickHouse → dashboard. Contact us if you want setup help.

​
Verify data is flowing

  1. Agent Analytics → Edge Traffic — overview and bot breakdown (Analytics Engine)
  2. High-signal events should reach ClickHouse within about a minute of test traffic
  3. For self-hosted: run the curl smoke test in the Cloudflare Worker guide
If overview counts work but ClickHouse-backed features do not, the ingest pipeline may be stalled — contact getcitable.com.

​
Common questions

Ingest expects classified edge events in the ObserveEvent format. In practice you need a Worker (managed or self-hosted) on the request path. Direct browser or server calls are not the intended integration.
No. When Connectors shows Active, Citable has already provisioned KV, the ingest token, and edge-proxy routing. You only add DNS.
Yes — contact us. We can migrate your siteId, provision managed routes, and coordinate token rotation so there is no gap in telemetry.
Disconnect the self-hosted connector in Connectors and create new credentials, or use Regenerate token when available. Update your Worker secret with npx wrangler secret put CITABLE_INGEST_TOKEN.
Managed: disconnecting in Connectors deletes KV entries and revokes the token. Self-hosted: contact us to revoke; remove or update the secret in your Worker.