Merge pull request #74 from zhongweili/fix/cron-firejob-and-llm-retry

fix(cron): include AccountID in fireJob InboundMessage + LLM retry

Author: idoubi

internal/agent/loop.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"crypto/sha256"
 	"encoding/json"
+	"errors"
 	"fmt"
 	"log/slog"
 	"os"
@@ -1776,6 +1777,50 @@ func (a *Agent) flushLeftoverSteer(sess *session.Session) {
 	}
 }
 
+// llmRetry wraps an LLM call with retry logic for transient errors (network
+// glitches, server 5xx, EOF). Context cancellation / deadline exceeded are
+// treated as terminal — there's no point retrying when the caller has gone
+// away or the deadline has passed. Uses exponential backoff (1s, 4s, 9s)
+// across up to wechatLLMRetryAttempts calls.
+//
+// The label argument is used for structured logging (typically a.name).
+const llmRetryAttempts = 3
+
+func llmRetry(ctx context.Context, label string, fn func(context.Context) (*provider.Response, error)) (*provider.Response, error) {
+	var lastErr error
+	for attempt := 1; attempt <= llmRetryAttempts; attempt++ {
+		resp, err := fn(ctx)
+		if err == nil {
+			if attempt > 1 {
+				slog.Info("LLM call succeeded after retries",
+					"agent", label, "attempts", attempt)
+			}
+			return resp, nil
+		}
+		lastErr = err
+
+		// Context errors are terminal — don't retry.
+		if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) {
+			return nil, err
+		}
+
+		if attempt < llmRetryAttempts {
+			backoff := time.Duration(attempt*attempt) * time.Second // 1s, 4s, 9s
+			slog.Warn("LLM call failed, retrying",
+				"agent", label, "attempt", attempt,
+				"max", llmRetryAttempts, "backoff", backoff, "error", err)
+			select {
+			case <-time.After(backoff):
+			case <-ctx.Done():
+				return nil, errors.Join(lastErr, ctx.Err())
+			}
+		}
+	}
+	slog.Error("LLM call failed after all retries",
+		"agent", label, "attempts", llmRetryAttempts, "error", lastErr)
+	return nil, lastErr
+}
+
 // HandleMessage processes an inbound message through the ReAct loop.
 func (a *Agent) HandleMessage(ctx context.Context, msg bus.InboundMessage) string {
 	// Check for slash commands first. Empty reply means "handled but
@@ -2039,14 +2084,16 @@ func (a *Agent) HandleMessage(ctx context.Context, msg bus.InboundMessage) strin
 			})
 		}
 		dumpLLMRequest(a.name, a.model, llmMessages, callTools)
-		resp, err := a.streamChatToResponse(ctx, llmMessages, callTools)
+		resp, err := llmRetry(ctx, a.name, func(ctx context.Context) (*provider.Response, error) {
+			return a.streamChatToResponse(ctx, llmMessages, callTools)
+		})
 
 		// Hook: AfterModelCall
 		hcAfter := &HookContext{AgentName: a.name, Point: AfterModelCall, Messages: messages, Response: resp, Error: err, StartTime: hcBefore.StartTime, Channel: msg.Channel, AccountID: msg.AccountID, ChatID: msg.ChatID, UserID: a.ownerUserID, GoalSessionKey: a.registry.GoalSessionKey()}
 		a.hooks.Run(ctx, hcAfter)
 
 		if err != nil {
-			slog.Error("LLM chat failed", "agent", a.name, "error", err)
+			slog.Error("LLM chat failed after retries", "agent", a.name, "error", err)
 			emitEvent(ctx, ChatEvent{Type: "error", Data: map[string]any{"message": err.Error()}})
 			emitEvent(ctx, ChatEvent{Type: "done"})
 			return "Sorry, I encountered an error processing your request."
@@ -2693,13 +2740,15 @@ func (a *Agent) HandleMessageStream(ctx context.Context, msg bus.InboundMessage)
 		a.hooks.Run(ctx, hcBefore)
 
 		dumpLLMRequest(a.name, a.model, messages, toolDefs)
-		resp, err := a.provider.Chat(ctx, messages, toolDefs, a.model, a.maxTokens, a.temperature)
+		resp, err := llmRetry(ctx, a.name, func(ctx context.Context) (*provider.Response, error) {
+			return a.provider.Chat(ctx, messages, toolDefs, a.model, a.maxTokens, a.temperature)
+		})
 
 		hcAfter := &HookContext{AgentName: a.name, Point: AfterModelCall, Messages: messages, Response: resp, Error: err, StartTime: hcBefore.StartTime, Channel: msg.Channel, AccountID: msg.AccountID, ChatID: msg.ChatID, UserID: a.ownerUserID, GoalSessionKey: a.registry.GoalSessionKey()}
 		a.hooks.Run(ctx, hcAfter)
 
 		if err != nil {
-			slog.Error("LLM chat failed", "agent", a.name, "error", err)
+			slog.Error("LLM chat failed after retries", "agent", a.name, "error", err)
 			return a.stringStream("Sorry, I encountered an error processing your request.")
 		}
 		a.meterTokens(ctx, sess.Key(), resp.Usage, 0)