What causes "agent error (no details)" in chenhg5/cc-connect?

Check agent process stderr/logs around the failure timestamp for the real cause

agent error (no details)

Error message

agent error (no details)

What it means

During relay event streaming, the agent emitted an EventError whose event.Error field was nil, so cc-connect returns a generic 'agent error (no details)' sentinel. The agent signalled failure without attaching an error object, leaving no cause to propagate. This is a defensive branch so callers always get a non-nil error.

Source

Thrown at core/engine.go:16072

			if currentID := agentSession.CurrentSessionID(); currentID != "" {
				saveRelaySessionID(currentID, true)
			}
			resp := event.Content
			if resp == "" && len(textParts) > 0 {
				resp = strings.Join(textParts, "")
			}
			if resp == "" {
				resp = "(empty response)"
			}
			slog.Info("relay: turn complete", "from", fromProject, "to", e.name, "response_len", len(resp))
			agentSession.Close()
			return resp, nil
		case EventError:
			agentSession.Close()
			if event.Error != nil {
				return "", event.Error
			}
			return "", fmt.Errorf("agent error (no details)")
		case EventPermissionRequest:
			// Auto-approve all permissions in relay mode
			_ = agentSession.RespondPermission(event.RequestID, PermissionResult{
				Behavior:     "allow",
				UpdatedInput: event.ToolInputRaw,
			})
		}
		if ctx.Err() != nil {
			// Relay timed out. Let the agent finish its turn in the
			// background so the session state is saved cleanly and the
			// session remains resumable for the next relay call.
			go e.drainRelaySession(agentSession, session, sessions, agent.Name(), relaySessionKey)
			return relayPartialResponseOrError(ctx.Err(), textParts, fromProject, e.name)
		}
	}

	// Event channel closed without EventResult.
	agentSession.Close()

View on GitHub (pinned to 4000b2338a)

Solutions

Check agent process stderr/logs around the failure timestamp for the real cause
Retry — intermittent agent crashes often include proper error detail on retry
Update or fix the agent adapter so it always populates EventError.Error
Check agent CLI version for known crash bugs and upgrade

Example fix

// before (adapter emitting the event)
events <- Event{Type: EventError}
// after
events <- Event{Type: EventError, Error: fmt.Errorf("agent exited: %s", stderrTail)}

Defensive patterns

Strategy: try-catch

Try / catch

case EventError:
    agentSession.Close()
    if event.Error != nil {
        return "", event.Error
    }
    slog.Error("agent error without detail; check agent stderr")
    return "", fmt.Errorf("agent error (no details)")

Prevention

Use first-party or well-tested agent adapters that always populate EventError.Error
Capture and log agent stderr so failures are attributable even when Error is nil
Keep agent CLIs updated to avoid unparseable crash output
Retry once on detail-less agent errors before surfacing to the user

When it happens

Trigger: The relay event loop receives event.Type == EventError with event.Error == nil — an agent adapter emitted a failure event without populating Error (adapter bug, or the agent CLI produced an unparseable failure output that was normalized to an empty error).

Common situations: Agent CLI crashed with output the adapter couldn't parse into a structured error; custom/third-party agent adapter with incomplete EventError handling; agent stderr lost due to pipe buffering so no detail was captured.

Related errors

AI-assisted analysis of chenhg5/cc-connect@4000b2338a (2026-09-06). Data as JSON: /api/errors/2dd67dbd5fba8243. Report an issue: GitHub.