Srinidhi P — February 14, 2026

Hey my MCP tools are registered and work when invoked manually, but A2A-triggered messages don’t result in tool execution or resolution. Is there a required config (agent mode, tool trust policy, or run trigger setting) to ensure `message/send` actually starts autonomous execution? Looking for the correct way to programmatically trigger a fully autonomous agent run from an external frontend?

Akash M — February 14, 2026

Hey team 👋 Here’s my hackathon submission: PitCrew Incidents (Archestra + MCP) : www.youtube.com/watch?v=OLO5L_fgJeY

Pixel22 — February 14, 2026

Getting issues with the quickstart guide, can anyone help, please?

8 replies

Sridhar Sing — February 14, 2026

I’ve created multiple agents, along with a parent agent that orchestrates them. When I send a request to the parent agent, it correctly delegates the task to a child agent. However, the request fails with an error stating that the *LLM API key is not set for the individual agent*. I’m confused because I’ve already configured the LLM API key globally. Despite this, the system seems to require the API key to be set separately for each agent. Is it actually necessary to configure an LLM API key for every individual agent, or should a single global key be sufficient?

Mohit Jeswani — February 14, 2026

i tried the gateway mode of archestra on first day it worked archestra_gateway_mode !! i think its allowed .. can i get a confirmation?

Samyak Bharsakle — February 14, 2026

hello organisers! sorry i had a busy week and now that the weekend is here im planning to hack something with Archestra for the 2 fast 2 mcp hackthon joining late wont disqualify me right? and just to double check I can make my submission till 15th Fe 11:59pm (GMT+5:30) right?

3 replies

Arjun — February 14, 2026

Hello! Had fun working on this. And AI said, Let there be Light. (a _light_ read, not ai-gen : D ) Cheers.

2 replies