What problem does it solve? Drawing a clear observability architecture diagram by hand is slow and error-prone: signal pipelines get merged, alert and backpressure paths are forgotten, and credentials leak into exported runbooks. This Skill produces a validated, left-to-right telemetry data-flow diagram in Excalidraw via the ExcaliDash MCP server. ## Core Features & Use Cases - Three-signal fan-out: Renders instrumented services feeding a collector/agent that splits into separate metrics, logs, and traces pipelines, each landing in its own storage backend (e.g. Prometheus, Loki, Tempo) with dashed telemetry edges and a signal legend. - Four distinct paths: Separates the happy path from the alert path (rule → notifier → on-call), the error/backpressure path (buffer full → drop/spool → dead-letter), and the on-call escalation chain (primary → secondary → incident). - Mandatory quality loop: Runs lint, score, and repair cycles until the drawing scores at least 95 with zero hard blockers, with rollback to checkpoints if a repair lowers the score. - Secret redaction: Replaces ingest tokens, remote-write credentials, and webhook URLs with typed [REDACTED_*] placeholders before any tool call and re-scans the export. - Use Case: Ask for a diagram of your OpenTelemetry pipeline where services emit to the collector, which fans out metrics to Prometheus, logs to Loki, and traces to Tempo, all visualized in Grafana with alerting and on-call escalation. ## Quick Start Ask the AI to diagram your observability pipeline, naming your services, collector, storage backends, and alerting tool, for example: draw our OpenTelemetry flow from services through the collector to Prometheus, Loki, and Tempo with Grafana dashboards and PagerDuty alerts.
