Question 1

Which errors does the auto-diagnose feature recognise?

Accepted Answer

Agata Stack currently pattern-matches for EADDRINUSE (port conflicts), module_not_found (missing Node dependencies), corrupted Next.js .next cache (routes-manifest.json missing), Prisma client not generated, Vite stale cache, permission denied errors, and out-of-memory (OOM) crashes. Hung processes are detected separately via CPU usage rather than log patterns.

Question 2

Does it run the fix command automatically, or do I have to approve it?

Accepted Answer

The fix command is surfaced as a runnable suggestion — you choose to run it. Stream rules can be configured to run fix commands automatically (for example, the built-in portInUse rule kills the port and restarts the service), but auto-run rules ship disabled for any action that modifies the filesystem and require explicit opt-in.

Question 3

How does it handle a service that is hung rather than crashed?

Accepted Answer

Agata Stack checks the CPU percentage of the managed process. If the process is at 80% or higher CPU while also failing health checks, it triggers a hung-process diagnostic rather than log-pattern analysis and flags the process as likely hung with a suggested restart.

Question 4

Can I access the diagnosis from a script or CI pipeline?

Accepted Answer

Yes. The Developer API exposes POST /diagnose/:name (in debug builds) and returns the full DiagnosticResult as JSON including summary, rootCause, suggestedFix, fixCommand, and the last 5 log lines as logContext. All manage and diagnostic endpoints are loopback-only for security.

Question 5

Does Agata Stack work with remote services, or only local ones?

Accepted Answer

Both. Remote services managed via SSH appear in the same dashboard with the same health checks and the same latency display. Log-pattern diagnosis applies to remote services because their logs are tailed locally via SSH — the same log lines are available to the diagnostic engine.

Auto-Diagnose: మీ సేవ విఫలమైనప్పుడు, Agata Stack ఎందుకు అని చెబుతుంది

Related Insights

క్లౌడ్ సేవ లేదా మరో టూల్ లేకుండా నిరంతర HTTP ఆరోగ్య తనిఖీలు

నా స్థానిక సేవలతో ఒకే menubar icon నుండి నా remote staging server ను నేను ఎలా నిర్వహిస్తున్నాను

నా సర్వీస్ 3amకి క్రాష్ అయింది, ఎవరికీ తెలియలేదు — మారింది ఇదే