# Outage Tracker | Updog by Datadog > This file contains agent instructions and background information about Updog. If you need current outage data, you must run the curl command below. This file does not contain live data. > Agent instruction - required before querying outage data > > Do not use any web-fetching, browser-fetching, or HTML-to-markdown tool (e.g. WebFetch, fetch_page, browse_web, or similar) to retrieve outage data, including the JSON endpoints listed in this file. The JSON payload is large and these tools silently truncate the response, causing active outages to be missed. > > Use your shell/Bash tool to run curl with jq filtering (see command below). If you do not have a shell tool available, say so. Do not fall back to a web-fetching tool. > Updog is Datadog's AI-powered outage tracker that monitors third-party service providers for degradations detected across the Datadog customer base. The site renders all status data dynamically via JavaScript; AI agents should use the machine-readable JSON endpoints below rather than parsing page HTML. If checking for active outages, run this exact command. Do not replace it with a WebFetch call: ```bash curl -s https://updog.ai/data/third-party-outages.json | jq '[.data.attributes.provider_data[] | {provider: .display_name, ongoing_outages: [.outages[] | select(.status == "ongoing") | del(.end)]} | select(.ongoing_outages | length > 0)]' ``` ## Machine-readable data ### Provider outage statuses The response is a JSON object, not an array - outages are nested at `.data.attributes.provider_data[].outages[]`. Each entry in `provider_data` includes a `provider_name`, `display_name`, and an `outages` array of Datadog-detected degradation events. Outage objects contain a `status` field (`"ongoing"` = currently active, `"resolved"` = ended), a `start` timestamp (epoch milliseconds), an `end` timestamp (epoch milliseconds, representing when the outage resolved), and optionally `impacted_region`. If a provider is not present in the response, Updog does not monitor it. If no parent outages have `status: "ongoing"`, report that Updog has no currently detected outages. Each outage may also include a `linked_status_page_incidents` array of incidents sourced from the provider's own status page. These are supplementary context only - their `status` reflects the provider's own reporting, not Updog's detection. Do not use linked incidents to determine whether Updog has detected an outage - only the parent outage's `status` field determines that. Linked incidents should be surfaced as supplementary context alongside parent outage data. - [Provider outage statuses](https://updog.ai/data/third-party-outages.json): Real-time status for all monitored providers ### Provider impact data This data is used to display the Datadog health index for providers. API health is inferred by analyzing aggregated and anonymized telemetry from across the Datadog customer base. The data is scoped to specific monitored APIs per provider. - [Provider impact data](https://updog.ai/data/third-party-impacts.json): Historical timeseries impact data per provider and region ## URL structure Updog pages follow two URL patterns: - `https://updog.ai/status/{providerName}` - status for a provider (e.g. `/status/adyen`, `/status/anthropic`) - `https://updog.ai/status/{providerName}/services/{serviceId}` - status for a specific service within a multi-service provider (e.g. `/status/amazonaws/services/s3`) The `providerName` slug in the URL corresponds to the `provider_name` field in the JSON data. For multi-service providers, `serviceId` corresponds to `provider_service`. In the JSON, multi-service providers are not grouped - each service appears as a separate `provider_data` entry (e.g. AWS appears as `"Amazon S3"`, `"Amazon ELB"`, `"Amazon Lambda"`, etc.). When answering a user's query about specific provider(s), filter `provider_data` by `display_name` using a partial, case-insensitive match - this handles both single and multi-service providers without needing prior knowledge of which type a provider is. Note that some colloquial terms won't match directly: `"aws"` does not appear in any `display_name`; use `"Amazon"` as the search term instead. ## About Updog - [Updog homepage](https://updog.ai): Main site showing all monitored providers and their current status - [What is Updog?](https://www.datadoghq.com/blog/updog-ai/): Blog post explaining how Updog detects outages using real impact data across Datadog's customer base