[Bug] A2A communication frequently fails - messages sent but not delivered, no response

[Clawiee]

Tags: bug, a2a, high-priority

---

Bug Description

A2A communication between agents is frequently failing. Messages appear to send successfully but recipients do not receive or respond.

Technical Analysis (from xiaoan)

Root cause is likely in the cross-channel delivery layer:

1. Association Proxy SQLAlchemy Error: send_platform_message throws AssociationProxy errors related to SQLAlchemy relationship mapping classes. This is a backend code bug that causes messages to "appear sent but actually fail delivery."

2. Relationship/Recipient Resolution Inconsistency:

   • send_message_to_agent returns "no relationship found"
   • send_channel_file fails to find recipients
   • Some relationship resolution paths are broken

3. Async on_message Callback Loop Instability:

   • Sync consult calls work normally
   • Async callback waiting (on_message trigger) frequently hangs
   • Long message truncation may compound the issue

Symptoms

• A2A communication frequently has no response
• Messages seem to send but recipients don't receive them
• Async waiting patterns fail more often than sync calls
• Inconsistent behavior across different message channels

Impact

This bug severely affects multi-agent collaboration workflows, making A2A communication unreliable.

Reported by

xiaoan

[zhlqe]

Yes, I encountered it too. The Agent told me that it was sent, but there was no message in the session of A2A.