The emerging standards behind AI agents: MCP, A2A, ACP and Open Responses each cover one boundary
bibryam · x · 2026-09-26
An article mapping the emerging standards of AI agent development: agents must connect to models, tools, remote agents, users, organization-specific knowledge and operational systems, and each boundary is forming its own contract — some are communication protocols, others APIs, telemetry vocabularies, instruction files or package formats.
- Open Responses covers the model connection layer: early integrations converged on Chat Completions concepts (messages, roles, streaming, structured tool calls), but modern agent runtimes need richer representations
- MCP is for tools and can be adopted without A2A (remote agents) or ACP (IDEs) — the standards are complementary, not bundled
- The piece positions each major standard within an agent architecture, serving as a practical selection map for agent engineering
More from coding & agent
- Codex merges Memory V2 prompts: 10KB summary cap, citations and selective retrieval — morqon · 2026-09-26
- AgentOdyssey, a Text-Game Engine for Test-Time Continual Learning Agents, Accepted at NeurIPS — DanielKhashabi · 2026-09-26
- Mario design rules for vertical agents: consistent inputs and progressively unlocked capabilities — dragon_khoi · 2026-09-26
- Agent UX is in 'Apple Newton' territory: why 90% accuracy loses to 100% — dragon_khoi · 2026-09-26
- Vertical agents: narrow task spaces create the '100% feeling' users need — dragon_khoi · 2026-09-26
- YC Decoded podcast: how coding agents are generalizing into robot-use agents — ycombinator · 2026-09-26