Search historical hosted-agent logs by time, severity, and text

Aperta
  • Qualcosa non va?
#10,922 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
35/100
Tipo di issue
Funzionalità
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
azure, github, go
Ambito
cli, observability

Direzione di ricerca

Read the released monitor implementation at cli/azd/extensions/azure.ai.agents/internal/cmd/monitor.go and the linked hosted-agent tracing setup to understand current log sources and permissions. The issue says an experience spec must first settle command placement, filter semantics, output, and data sources, so implementation needs design work before a code entry point is clear. Done means the specified workflow covers retained data limits, filtering, isolation, output, failures, help, and examples.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

area/telemetry enhancement ext-agents ext-foundry

Customer problem

An operator needs to answer questions such as "Which errors occurred for this deployed agent in the last hour?" or "Where did this timeout appear across sessions?" They should not need the original application checkout, a live session, or a hand-written telemetry query for these common investigations.

Current behavior

In released azure.ai.agents 1.1.0, azd ai agent monitor targets one hosted-agent session and supports console/system selection, a trailing line count of 1-300, and following new entries. It has no time-window, severity, or search-pattern filters.

Existing example:

azd ai agent monitor support --session-id "<session-id>" --type console --tail 100

Azure CLI and Application Insights/KQL can query retained application telemetry today. That workaround requires identifying the telemetry resource, applying the right agent filter, and carrying context between tools. Application Insights application logs are not guaranteed to contain every hosted-agent console line or system event.

Requested experience

Add an agent-scoped historical log-query workflow to the Agents extension, using available retained telemetry rather than treating a live session tail as a historical search.

Illustrative command shape, not an available command:

azd ai agent logs query --project-endpoint "$Project" --agent-name support --since 1h --level error --output json --no-prompt

The experience spec should settle command placement, time and text-filter semantics, output, and supported data sources before implementation.

Acceptance criteria

  • Query a bounded relative or absolute time window for a known remote agent, with severity and text filtering and optional version/session narrowing where the data supports them.
  • Support investigations across retained sessions without requiring azure.yaml or a currently running session.
  • Carry the selected Foundry project and deployed-agent context into the query. Resolve connected telemetry where possible and provide explicit, actionable guidance when extra configuration is needed.
  • Identify supported retained data sources and coverage. Do not assume the Application Insights traces table contains all console/system logs or promise reconstruction of data that was never collected.
  • Distinguish no matching records from missing telemetry configuration, unsupported data sources, and access failures. Explain relevant retention/sampling limits.
  • Provide useful terminal output and stable JSON with timestamps and available agent/session/trace correlation. Define result limits and pagination.
  • Preserve current monitor/follow behavior, reject incompatible selectors, and cover cross-agent isolation, time/filter behavior, and non-interactive failures. Update help and examples.

Related work and evidence

Lingua principale
Go
Stelle
573
Fork
387
Merge medio
2g 20h
PR unite (30g)
196

Preparare l'ambiente

Apri in Codespaces

Avvia il container di sviluppo del progetto nel browser, con il tuo account GitHub.

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di Azure/azure-dev

Tutte le issue di Azure/azure-dev

Issue simili

Altre issue su Go

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.