Agent skill · data analytics · jeremylongshore
ideogram-observability
Instrument Ideogram request, queue, async, safety, storage, latency, and spend signals without logging image content. Use when building dashboards, alerts, or SLO evidence. Trigger with "monitor Ideogram", "add Ideogram metrics", or "audit Ideogram telemetry".
Why this skill is useful
Provides specific metrics and monitoring scripts for Ideogram API that are not commonly known or documented.
What it needs
Requires typescript installed locally. Requires ideogram account access. About 2k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Ideogram Observability Overview Measure whether accepted image work becomes a safe durable asset within its deadline and budget. Join application request, queue, vendor, webhook or polling, download, storage, and publication stages using opaque identifiers rather than prompts or media. Prerequisites Service boundaries, SLO, error budget, budget window, owners, and incident thresholds. A data dictionary and cardinality policy for metrics, logs, and traces. Known endpoint, queue, async state, safety, storage, and cleanup transitions. Current Contract Useful vendor-facing signals include endpoint family, HTTP status, generationid, async state, isimagesafe, URL-present boolean, and timing. Default capacity is 10 in-flight requests, while actual account behavior and current pricing must be observed at decision time. Authentication Telemetry must record only that server-side Api-Key authentication was configured or rejected. Never capture the header value, prompts, structured prompts, source images, output images, signed URLs, or raw payload bodies. Instructions 1. Define the user outcome as a safe approved asset durably stored by its deadline, not merely HTTP 200. 2. Instrument admission, queue wait and age, in-flight count, upload, generation, webhook, polling, download, validation, storage, review, and cleanup stages. 3. Count statuses by endpoint and environment; separate auth, validation, throttling, capacity, safety, expiry, and storage failures. 4. Track accepted async generations to one terminal state and alert on stuck, duplicate, missing-webhook, or unreconciled records. 5. Attribute estimated or reconciled spend to useful, unsafe, failed, abandoned, and unstored outcomes. 6. Build SLO dashboards and actionable alerts with owners, runbook links, burn windows, and content-free exemplars. 7. Test alert delivery and verify telemetry deletion and access controls. Tool Discipline Use Read, Glob, and Grep to inspect instrumentation, dashboards, and logs. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/ideogram-observability