Skip to content

feat: Add LLM A2A Integration module#215

Open
hungpq-ai wants to merge 2 commits into
apexive:18.0from
hungpq-ai:feature/llm-a2a-integration
Open

feat: Add LLM A2A Integration module#215
hungpq-ai wants to merge 2 commits into
apexive:18.0from
hungpq-ai:feature/llm-a2a-integration

Conversation

@hungpq-ai

Copy link
Copy Markdown

Add Agent-to-Agent (A2A) protocol integration for LLM, enabling communication between AI agents using Google's A2A protocol.

Features:

  • Connect to external A2A agents
  • Delegate tasks to specialized agents
  • Agent discovery via Agent Cards
  • Streaming support for real-time responses
  • Multi-agent orchestration within LLM chat

Author by Hung Pham Quang (VietNam)

Add Agent-to-Agent (A2A) protocol integration for LLM, enabling communication
between AI agents using Google's A2A protocol.

Features:
- Connect to external A2A agents
- Delegate tasks to specialized agents
- Agent discovery via Agent Cards
- Streaming support for real-time responses
- Multi-agent orchestration within LLM chat

Author by Hung Pham Quang (VietNam)
@adar2378

Copy link
Copy Markdown
Contributor

@hungpq-ai Thanks for the contribution, we will check and get back to you.

@adar2378 adar2378 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@hungpq-ai

Testing Feedback

Thanks for this module! I tested it with the Google ADK hello_world_agent and found an issue.

Setup ✅

  • A2A agent running at http://localhost:9999
  • Created llm.a2a.agent record, test connection successful
  • Assigned agent to thread (GPT-4o)

Issue ❌

Agent delegation doesn't work - LLM acknowledges it will use the agent ("I'll ask the agent...") but never makes the actual tool call. No HTTP request reaches the A2A agent server.

Logs:

INFO: Added A2A capability prompt: 1 agents
INFO: Added 1 A2A delegation tools to thread 1
INFO: Storing 1 A2A tools for provider 1
INFO: POST https://api.openai.com/v1/chat/completions "HTTP/1.1 200 OK"
# No tool call, no request to A2A agent ❌

Questions

  1. Have you tested end-to-end delegation successfully?
  2. Which LLM model did you test with?

Suggestion

Add a README with:

  • Quick start guide (how to set up an A2A agent)
  • Example agent configuration
  • Expected behavior

Test env: Odoo 18.0, llm_openai, gpt-4o, Google ADK v1.21.0

@hungpq-ai

Copy link
Copy Markdown
Author

Okay, I’ll get back to you soon. I’m a bit busy at work right now.

@ayushin

ayushin commented Jan 20, 2026

Copy link
Copy Markdown
Contributor

@hungpq-ai looking forward to the fixed mergeable code

- Added llm_openai as a dependency to ensure proper method resolution order.
- Introduced a comprehensive README.md detailing features, installation, quick start, and troubleshooting for the LLM A2A integration.
@hungpq-ai

Copy link
Copy Markdown
Author

@hungpq-ai looking forward to the fixed mergeable code

I’ve updated my commit. Please have a look and let me know if you find any issues.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants