AEP: Agent Event Protocol — open async event layer for agents, tools, memory, and multi-agent runtimes. 4-language reference with cross-language conformance.
Agent Event Protocol (AEP) is a proposed open protocol for asynchronous communication between large language model agents, tools, memory systems, context providers, environment observers, and multi-agent runtimes.
AEP is designed as the asynchronous counterpart to MCP. MCP is good at synchronous capability invocation: listing tools, calling tools, reading resources, and returning immediate results. AEP focuses on the communication patterns MCP does not naturally cover: event streams, long-running task lifecycle, background feedback, memory updates, context invalidation, durable delivery, replay, cancellation, and agent-to-agent coordination.
AEP 0.1 draft is a multi-language protocol repository with:
- 17 protocol specifications covering session, subscription, task, error, versioning, delivery, reliability, security, conformance, and transport layers - 4 reference implementations (TypeScript, Python, Go, Java) with cross-language conformance verified against shared fixtures - 4 transport bindings (stdio, WebSocket, SSE, gRPC) implemented across all languages - SQLite-backed delivery stores with retry, dead-letter, and replay support in every language - ~320 tests across four languages, all passing - Spec site at axisrobo.github.io/aep
Agents should not only ask for capabilities. They also need to listen, react, coordinate, and recover.
AEP defines a common event model so that agents can receive asynchronous feedback from tools, other agents, memory systems, external environments, and runtime infrastructure without forcing every system to invent a custom callback, polli
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/axisrobo--aep.html)
AI Governance Infrastructure — open gateway demo. Self-hosted systemprompt.io gateway in a single Rust binary: policy, audit trails, and cost attribution on every AI request, with the Systemprompt Bri
neuronto/agentic-resource-discoveryNeuronto Agentic Resource Discovery (ARD) Index. Federated search across every public ARD registry, plus a verified tool index read from each MCP server's own tools/list, hybrid lexical and semantic r
sandbaseai/deepseek-harness-handbookIndependent, source-backed handbook for DeepSeek AI's official DeepSeek Harness (dsh): agents, plugins, security, troubleshooting, and runbooks.
AIYantra/extraThe Open-Source Astra 6 for your PC. Your AI can see, click, type, navigate, and get real work done on Windows.
shitianfang/jev-useClaude Code / Codex / pi plugin that hands agent steps needing no text output to Jev (TypeSafe's judgment model) — measured p50 ~230 ms and ~$0.02 per 1,000 judgments, with typed escalation back to th
pairbrowse/pairbrowseMultiplayer browser sessions for developers and AI coding agents
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.