Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.0xarchive.io/llms.txt

Use this file to discover all available pages before exploring further.

Status monitoring turns API health, ingestion state, freshness, incidents, latency, and SLA signals into a decision about whether downstream work should run. The public status page is a product surface. It checks service health, coverage, incidents, and market-family signals. The docs surface explains how those signals should be used by clients.

Status Signals

SignalUse it forDocs
REST livenessWhether the API edge is respondingREST API
WebSocket livenessWhether stream service checks are healthyWebSocket
CoverageWhich venue families, symbols, and data families are availableData Availability
IncidentsKnown interruptions or degraded windowsData Gaps
FreshnessWhether latest data is inside toleranceData Quality
Request IDsSupport and debugging across failed probesResponses, Errors

Client Policy

1

Separate service health from data trust

A healthy service can still return data that is stale for a specific job.
2

Check the relevant family

Inspect the venue family, symbol, and data family your workflow needs, not only global status.
3

Record the decision

Store status, coverage, incident, freshness, and request-ID context with downstream output.
4

Fail visibly

If status or freshness is outside tolerance, stop or mark the output instead of silently continuing.

Status Check Packet

Record enough context to distinguish a global outage from a stale market, a coverage miss, or a client-side request problem.
FieldCapture
Check typePublic status page, /v1/data-quality/status, coverage, incident, freshness, latency, or SLA
Route and methodThe exact API route or status surface the workflow checked
Venue and marketHyperliquid core, Spot, HIP-3, HIP-4, Lighter, symbol, pair, or market slug
Data familyTrades, candles, L2, L3, L4, funding, open interest, liquidations, exports, or streams
Observed timeThe UTC time the check ran and the latest timestamp returned by the data surface
ToleranceThe freshness or completeness threshold the downstream workflow requires
Request contextStatus code, error code, request_id, cursor, and retry count when available
DecisionContinue, delay, narrow the request, mark the output, or escalate with support context
Use the global status view to decide whether a broad service issue is already known. Use symbol-level coverage and freshness checks to decide whether one workflow should run right now. Those are related signals, but they should not collapse into the same client decision.

Run Or Delay Rule

Status checks should end in one of a few explicit decisions.
DecisionUse whenClient behavior
ContinueService health, coverage, and freshness are inside the workflow toleranceRun the job and store the check result with the output
NarrowOne venue family, symbol, schema, or time window is outside toleranceReduce scope and record the excluded market or range
DelayThe needed data is stale, incident-affected, or still catching upStop the job and retry after the next scheduled check
Mark outputThe workflow can tolerate partial context but the reader must see itAttach the incident, freshness, or coverage state to the output
EscalateThe issue blocks a customer workflow or repeats across checksSend route, symbol, timestamps, request ID, and status context to support
Use the same rule for REST pulls, WebSocket replay, SDK reconstruction, Data Catalog exports, dashboards, alerts, and model inputs. The point is to keep a stale market or incident window from looking like a normal result. Use Data Quality for route-level quality checks, Reliability And Gaps for workflow design, and WebSocket Keep Alive for stream recovery.

Escalation Notes

When a status check affects a customer workflow, include the route, symbol, data family, observed timestamp, expected freshness tolerance, request ID, and whether the issue is global or market-specific. That keeps support from treating a stale Lighter stream, a HIP-3 gap, and a general API outage as the same incident.
Last modified on May 18, 2026