Hub
Ops / observability hub
Everything you need to monitor, trace, and debug Meridian in production. Pick your stack, wire up the signals, and keep your workloads visible.
Observability Tools
Overview of the observability stack powering Meridian — traces, metrics, logs, and the signals that matter.
With Langfuse
Open-source LLM tracing. Capture every completion, embedding, and tool call with sub-millisecond overhead.
With Helicone
Lightweight proxy for logging, caching, and rate-limiting LLM requests. Drop-in integration with zero code changes.
With LangSmith
Full-cycle LLM observability — debug traces, run evals, and monitor regressions across model versions.
With Braintrust Proxy
Eval-first observability. Ship prompts, collect scores, and iterate with a unified proxy layer.
With Arize
Production ML monitoring. Drift detection, performance tracing, and root-cause analysis for deployed models.
Logging & Retention
Structured JSON logging, log levels, retention policies, and export pipelines to your own storage.
Error Codes
Complete reference of Meridian error codes — HTTP status, internal codes, and human-readable explanations.
Request ID
How Meridian generates, propagates, and surfaces the x-request-id header for end-to-end traceability.
Status
Live system status, incident history, and uptime SLA reporting for the Meridian platform.
Pro tip: Combine Langfuse traces with Meridian request IDs to correlate frontend latency with backend model calls. See Request ID for the integration guide.