[dotnet-port-api] BackgroundTaskCompletion loop evaluator is blocked on a missing subagent/background-agents harness

Aperta
#1,164 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
20/100
Tipo di issue
Funzionalità
Chiarezza
Specificata chiaramente
Stato di attività
Attiva
Stack tecnologico
go

Direzione di ricerca

The issue is blocked on designing and implementing a missing subagent/background-agents harness in the Go SDK. Start by reviewing the .NET implementation referenced, the parity doc (docs/dotnet-go-sdk-feature-comparison.md), and the existing agent/, tool/, and workflow/ directories to understand the gap. A design proposal for the BackgroundAgentsProvider and BackgroundTaskInfo equivalents is needed before any code can be written.

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

Descrizione

While porting the loop evaluators (see #1162, #1163), I found that BackgroundTaskCompletionLoopEvaluator cannot be ported yet — its entire dependency does not exist in the Go SDK.

The blocker

The .NET BackgroundTaskCompletionLoopEvaluator keeps re-invoking the agent until a BackgroundAgentsProvider reports no still-running background tasks, formatting the running BackgroundTaskInfo list ({incomplete_tasks} / {incomplete_task_count}) into the feedback.

Go has none of that layer:

  • No BackgroundAgentsProvider, BackgroundTaskInfo, or any subagent/background-task harness (a grep for backgroundagent/subagent across agent/, tool/, workflow/ returns nothing).
  • The parity doc (docs/dotnet-go-sdk-feature-comparison.md) confirms it: the harness row lists "subagent harness utilities" among what Go "still lacks," and the top-level gaps list includes "subagents."

So the evaluator is a thin consumer of a subsystem that first has to exist.

Proposed sequencing

  1. Design + port the subagent / background-agents harness (a BackgroundAgentsProvider-equivalent context provider that spawns and tracks background agent runs, plus a BackgroundTaskInfo-equivalent). This is a sizeable owner-play feature and deserves its own design discussion — happy to draft an architecture proposal (Mermaid + API sketch) if there is interest.
  2. Once that exists, BackgroundTaskCompletionLoopEvaluator is a small follow-up (same shape as the AI-judge/todo evaluators), subject to the same placement question raised in #1163.

Flagging rather than implementing, since building the whole subagent subsystem to satisfy one evaluator would be a large, design-first change — not a drive-by PR.

Lingua principale
Go
Stelle
659
Fork
67
Merge medio
18h 49m
PR unite (30g)
81

Preparare l'ambiente

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 microsoft/agent-framework-go

Tutte le issue di microsoft/agent-framework-go

Issue simili

Altre issue su Go

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.