How should AI agents actually discover and talk to each other?
RaraAvis27 · reddit · 2026-09-13
A developer asks how agents in multi-agent systems should discover and communicate with each other, listing current approaches: hardcoded calls, a central orchestrator, MCP/tool-based integration, A2A-style protocols, message queues, and agent directories. The thread invites practitioners to share what they actually ship and what breaks first once systems grow beyond 2-3 agents.
More from coding & agent
- Dev goes 'voice pilled' with Wispr Flow, dictating 15-minute prompts to his agent — fernandorojo · 2026-09-13
- Labeler Agreement Check: If Two Trained Labelers Can't Agree, the Rubric Is the Bug — blaizedsouza · 2026-09-13
- Learn Omarchy: interactive desktop course with 17 lessons and 101 activities for DHH's distro — DanWahlin · 2026-09-13
- Better than /goal: orchestrate coding subagents with a top model and integration criteria — gregmushen · 2026-09-13
- Dev shares 3 personal AI apps he uses daily: a 24/7 voice desktop agent and a GPUI Messages rewrite — johnlindquist · 2026-09-13
- Adopt one tip at a time: fix your current annoyance, then bake it into your instructions — TheMoonMidas · 2026-09-13