DeltaFeedsV0

Context collection · active

Agent Operations

Operational references for observing agent workloads, recovering long-running work and checking service health.

observabilityoperationsreliability

Collection updated 2026-11-08T22:15:37.133322Z

Retrieve context JSON →Checkpoint documentation

Context items

OpenTelemetry GenAI semantic conventions

OpenTelemetry maintains a dedicated repository for generative-AI telemetry conventions, including spans, metrics and events for model clients, MCP and provider-specific behavior. The conventions extend the core OpenTelemetry vocabulary. The repository is the current destination linked from the former GenAI documentation page.

OpenTelemetry project repository · Updated 2026-11-08T22:15:37.327641Z

External source ↗

Prometheus: instrumenting services and batch jobs

Prometheus instrumentation guidance distinguishes online services, offline processing and batch jobs. For online services it highlights request volume, errors, latency and in-progress work. These categories help organize measurements for an agent service without treating every individual request as a human-facing alert.

Prometheus project documentation · Updated 2026-11-08T22:15:37.497994Z

External source ↗

Temporal: durable workflow execution and replay

Temporal explains how workflow executions recover progress using persisted event history and replay. Its documentation distinguishes the workflow definition from a running execution and describes states such as completed, failed and cancelled. These concepts are relevant to agents coordinating work across multiple external calls.

Temporal documentation · Updated 2026-11-08T22:15:37.661917Z

External source ↗

Kubernetes: liveness, readiness and startup probes

Kubernetes separates probes for whether a container should restart, whether it can receive service traffic and whether startup has completed. The documentation shows how startup probes protect slow initialization and how readiness can temporarily remove an unready workload from service traffic.

Kubernetes documentation · Updated 2026-11-08T22:15:37.838457Z

External source ↗

Docker HEALTHCHECK: application health inside a container

The Dockerfile HEALTHCHECK instruction runs a command to inspect application health inside a container. Docker tracks a health status separately from whether the container process is running. The reference explains intervals, timeouts and failure handling for defining that check.

Docker documentation · Updated 2026-11-08T22:15:37.997250Z

External source ↗