Supabase Health Check Advisors API for Automated Service Error Detection

Overview

Supabase Health Check Advisors API for Automated Service Error Detection Supabase has made Health Check Advisors generally available, giving developers a read-only way to detect persistent recent 5xx error rates across key services. The Supabase Health Check Advisors API can turn that signal into an automated post-deployment check, helping students, freelancers, and engineering teams identify where to begin investigating a failure. AreaWhat is confirmedPractical meaningCoverageData API, Auth, Storage, and Edge FunctionsNarrows investigation to a serviceAccessSupabase Studio or Management APISupports manual and programmatic checksResultSeverity levels, empty results, and unavailable-check statusUseful for routing triage workAvailabilityGeneral Availability; self-hosted status is N/AApplies to supported hosted projects

What Supabase Health Check Advisors changed and why it matters

In its September 18, 2026 changelog announcement, Supabase described Health Check Advisors as generally available. Version one focuses on service error-rate checks rather than attempting to diagnose every infrastructure condition. The official feature page identifies the supported services and explains that checks use recent request logs. The four v1 checks The advisors look for persistent elevated 5xx rates in the Data API, Auth, Storage, and Edge Functions. Results include severity information. An empty result means the checks ran without finding a high error rate, while advisor_check_unavailable indicates that a check could not run.

Why an agent can use the result

An automation workflow can treat each finding as a routing signal: identify the affected service, save the result, and direct a person or debugging agent toward the relevant logs. This is a useful pattern to practise alongside an application development project. Database connectivity probes and instance-health checks are described as future roadmap items, so version one has a deliberately narrower scope.

Why a service error needs a narrower starting point

A generic HTTP 5xx response says that a server-side failure occurred. It does not identify whether the Data API, Auth, Storage, or Edge Functions produced the error. That distinction matters because each service has different logs, configuration, dependencies, and likely investigation paths. From symptom to service Health Check Advisors reduces this first search. It examines recent request logs and reports persistent elevated error rates for supported Supabase services. A developer can use the affected service and severity as a routing signal, then inspect the corresponding logs and recent application changes. Supabase’s Advisors documentation points developers toward logs when they need to continue troubleshooting. This creates a useful learning project. A student can practise the same pattern in a software application development project: capture an error, identify its service boundary, record the advisor result, and open the relevant evidence for review. The goal is not to build an automatic fixer. The goal is to shorten the path from a broad symptom to a focused investigation.

What the result does not prove

A finding is a starting signal, not proof of root cause. A high 5xx rate may reveal where failures are concentrated, while the underlying cause could involve application code, authentication settings, storage requests, deployment changes, or another dependency. The Health Check Advisors feature description frames the checks as a way to identify service-level error-rate problems, not as a complete outage monitor. Use the result to choose the next diagnostic question: which service shows the signal, which request pattern changed, and what do its logs show? This keeps automated triage useful without confusing a health indicator with a complete explanation.

How it works

Supabase Health Check Advisors API for automated service error detection: access and limits Health Check Advisors is generally available for supported hosted Supabase projects. Supabase marks self-hosted availability as N/A. Teams can run checks in Studio, while automated workflows use the Management API endpoint POST /v2/projects/{ref}/advisors/run. The official endpoint reference defines the project reference path, request requirements, response structure, and supported HTTP responses. Who can call the endpoint The request needs authenticated HTTPS access and a fine-grained token with the advisors_read permission. This read-only scope fits a post-deployment checker that records findings without changing project configuration. Store the token on a server, never in browser code or client logs. A learner can practise this pattern while building a software project consultation workflow that separates monitoring credentials from user-facing features. Successful responses can contain advisor findings with severity and check information. Authentication failures can return HTTP 401, permission failures can return HTTP 403, and excessive request volume can return HTTP 429. The endpoint reference also documents HTTP 200 for a successful request. Availability and request limits Supabase’s Management API documentation lists a standard limit of 120 requests per minute per user, project, organization, and endpoint. Rate-limit headers help a client track remaining capacity. A scheduled checker should handle HTTP 429 responses instead of retrying continuously. Supabase billing uses Free, Pro, Team, and Enterprise plans, with fixed subscription fees and variable usage fees described in its billing documentation. No feature-specific Health Check Advisors price is stated on the cited product pages. Management API access remains subject to Supabase’s authentication, rate-limit, and plan rules.

How to build an automated service-error check

The Management API turns Health Check Advisors into a repeatable post-deployment check. A server-side workflow sends an authenticated request for a project, receives the advisor snapshot, and stores the response with the deployment time. The official Health Check Advisors announcement documents the endpoint and its supported result states. The request and response The request uses the project reference in the URL and the POST method. Supabase’s official example follows this pattern: curl --request POST 'https://api.supabase.com/v2/projects/{ref}/advisors/run' \ --header 'Authorization: Bearer YOUR_TOKEN' \ --header 'Content-Type: application/json' \ --data '{}' The Management API reference describes the response structure, including advisor check identifiers and severity values. An empty result means the checks completed without finding a high error rate. The advisor_check_unavailable result identifies a check that could not run. Findings should be classified before they are sent to a person, ticketing system, or AI workflow development service. A safe automation flow Keep the workflow small and observable. First, trigger it after a deployment or on a schedule. Next, call the endpoint once and record the HTTP status, timestamp, project reference, and returned findings. Treat HTTP 200 as a completed request, then separate elevated findings, empty results, and unavailable checks. Handle HTTP 401, 403, and 429 as request or access conditions that need their own operational response. Finally, route an elevated result to the matching service logs for investigation. The feature uses cached snapshots, so repeated calls may return the same result and a finding may take time to clear after a fix. The API narrows the first search; it does not perform the repair.

Practical steps

Testing the signal after deployment A useful learning exercise is to treat Health Check Advisors as a routing signal during a controlled release. Change one service or request path, run the checker after deployment, and record the returned check identifier, severity, timestamp, and deployment version. Then compare that snapshot with the relevant service logs. Supabase describes Advisors as deterministic checks based on recent project activity, while its troubleshooting guidance directs developers toward logs for deeper investigation. A practical post-deployment exercise Start with a read-only report. The checker should save each response in a small history store, such as a database table or JSON file, so learners can compare snapshots across releases. Sort elevated findings above empty results, and send unavailable checks to a separate review path. This creates a realistic portfolio project for anyone practising application development services: the result is an observable workflow, not a dashboard that claims to explain every failure. For evaluation, ask three questions after each run: did the signal identify the expected Supabase service, did the matching logs show related 5xx responses, and did the finding change after the measurement window moved? An empty response means the advisor did not find a high error rate. It does not prove that application behavior, latency, client-side errors, or every infrastructure condition is healthy. Health Advisor versus Health agent Do not combine two similarly named Supabase features. Health Check Advisors expose service error-rate checks through Studio and the Management API. Supabase separately documents a read-only Health agent workflow that can monitor hourly, run after deployment, inspect API and Auth errors, and examine Postgres connection pressure through its documented agent setup. The distinction matters for project design. Use Advisors when your automation needs a structured API result that narrows investigation to a service. Consider the Health agent workflow when your team wants a broader, read-only monitoring prompt with agent-based review. Neither removes the need to inspect evidence before deciding what to change.

When the Supabase Health Check Advisors API Is Worth Using

Use the Supabase Health Check Advisors API for automated service error detection when your team needs a repeatable, read-only signal that narrows investigation to a Supabase service with elevated recent 5xx errors. This decision fits student projects, freelance applications, and production teams that already have deployment workflows but need a clear first debugging target. The API is most valuable when a deployment checker can connect an advisor result to useful operational context. Store the project reference, deployment identifier, timestamp, HTTP response status, check identifier, severity, and investigation link together. This record gives a learner a practical example of structured API handling and gives a team a small evidence trail for comparing releases. Make one next step: create a server-side post-deployment check that stores the advisor snapshot and opens the relevant service logs for human or agent review. The checker should call POST /v2/projects/{ref}/advisors/run with a least-privilege Management API token, then route the returned finding to the matching Data API, Auth, Storage, or Edge Functions logs. Keep the token in server-side secret storage. Never place it in browser code or public logs. A strong first version needs no elaborate dashboard. A scheduled job, a small history table, and a clear record of each response are enough to practise authentication, HTTP status handling, structured responses, and incident triage. After that foundation works, add notification rules or an agent review step based on the team’s needs. The decision is practical: choose Health Check Advisors when narrowing service-level error investigation matters more than building a broad monitoring platform. Start with one project, one post-deployment trigger, and one stored snapshot per run.

Additional image