docs: add Atmosphere framework to agentic AI frameworks

Author: jfarcand

educational-resource/docs/30-agentic-ai/20-frameworks/50-atmosphere.md

@@ -0,0 +1,117 @@
+---
+title: Atmosphere
+sidebar_position: 50
+hide_title: true
+
+level: intermediate
+status: published
+visibility: public
+
+external-link: https://github.com/Atmosphere/atmosphere
+programming-language: "Java"
+---
+
+# Atmosphere
+
+**<h2><a href="https://github.com/Atmosphere/atmosphere" target="_blank" rel="noopener noreferrer">👉 Visit Atmosphere on GitHub ↗</a></h2>**
+
+[Documentation](https://atmosphere.github.io/docs/) · [Tutorial](https://atmosphere.github.io/docs/tutorial/01-introduction/) · [Samples](https://github.com/Atmosphere/atmosphere/tree/main/samples)
+
+Atmosphere is a real-time transport abstraction framework for Java. It handles WebSocket, SSE, HTTP long-polling, and gRPC — with automatic transport negotiation, message caching, reconnection, and clustering. Built and maintained over many years.
+
+The AI agent layer was built on top of that foundation. An `@Agent` class defines prompts, commands, and tools. The AI execution engine — tool calling, memory, RAG, retries — is pluggable via the `AgentRuntime` SPI and determined by classpath: Spring AI, LangChain4j, Google ADK, or Embabel. The framework also handles protocol exposure (MCP, A2A, AG-UI) and messaging channels (Slack, Telegram, Discord).
+
+## Architecture
+
+Atmosphere follows the Servlet model: application code is separate from the execution engine.
+
+| Servlet analogy | Atmosphere equivalent |
+|-----------------|----------------------|
+| `Servlet` / `@Controller` | `@Agent` / `@Prompt` |
+| Tomcat, Jetty, Undertow | LangChain4j, Spring AI, ADK, Embabel |
+| `AsyncSupport` (container SPI) | `AgentRuntime` (framework SPI) |
+| HTTP, WebSocket | MCP, A2A, AG-UI |
+| `web.xml` / annotations | Skill files (Markdown + YAML) |
+
+The `@Agent` class declares what the agent does. The `AgentRuntime` on the classpath determines how it executes: tool calling loops, conversation memory, RAG retrieval, retries, and guardrails.
+
+## Supported Runtimes
+
+| Runtime | Module | Capabilities |
+|---------|--------|-------------|
+| **Built-in** | `atmosphere-ai` | OpenAI-compatible client (Gemini, OpenAI, Ollama) |
+| **LangChain4j** | `atmosphere-langchain4j` | ReAct tool loops, `StreamingChatModel`, retries. `@AiTool` methods bridged to LangChain4j tools. |
+| **Spring AI** | `atmosphere-spring-ai` | `ChatClient`, function calling, RAG advisors |
+| **Google ADK** | `atmosphere-adk` | `LlmAgent`, function tools, session management |
+| **Embabel** | `atmosphere-embabel` | Goal-driven GOAP planning (experimental) |
+
+Switching runtimes requires changing one Maven dependency. Agent code, tools, commands, and skill files remain the same.
+
+## Framework Capabilities
+
+These work regardless of which `AgentRuntime` is active:
+
+- **Real-time streaming** — LLM tokens delivered via WebSocket/SSE
+- **Protocol exposure** — MCP, A2A, and AG-UI endpoints auto-register from classpath
+- **Multi-agent orchestration** — `@Coordinator` + `@Fleet` for parallel fan-out, sequential pipelines, coordination journal
+- **Conversation memory** — multi-turn context persisted to SQLite or Redis
+- **Tool portability** — `@AiTool` methods work across all runtimes
+- **Skill files** — Markdown with YAML frontmatter for agent persona, tools, guardrails, and channel routing
+- **Messaging channels** — Slack, Telegram, Discord, WhatsApp, Messenger via bot token configuration
+- **Servlet integration** — runs as a filter in Spring Boot, Quarkus, or any Servlet 6.0+ container
+
+## Quick Start
+
+The Atmosphere CLI scaffolds new projects and runs existing samples directly:
+
+```bash
+brew install Atmosphere/tap/atmosphere
+
+# Scaffold a new project
+atmosphere new my-agent --template ai-chat
+cd my-agent && LLM_API_KEY=your-key ./mvnw spring-boot:run
+
+# Or run any of the 18 built-in samples
+atmosphere list
+atmosphere run spring-boot-ai-chat
+```
+
+```java
+@Agent(name = "my-agent", description = "What this agent does")
+public class MyAgent {
+
+    @Prompt
+    public void onMessage(String message, StreamingSession session) {
+        session.stream(message);  // Dispatches to active AgentRuntime
+    }
+
+    @Command(value = "/status", description = "Show status")
+    public String status() {
+        return "All systems operational";
+    }
+
+    @AiTool(name = "lookup", description = "Look up data")
+    public String lookup(@Param("query") String query) {
+        return "Result for: " + query;
+    }
+}
+```
+
+## Protocol Support
+
+| Protocol | Standard |
+|----------|----------|
+| MCP | Model Context Protocol |
+| A2A | Agent-to-Agent Protocol |
+| AG-UI | Agent-User Interaction Protocol |
+
+Protocols auto-register based on classpath detection.
+
+## Requirements
+
+Java 21+. Spring Boot 4.0, Quarkus 3.21, or any Servlet 6.0+ container.
+
+## Related Resources
+
+- [A2A Protocol](./40-a2a-protocol.md) — Atmosphere implements A2A for agent-to-agent communication
+- [Google ADK](./10-adk.md) — supported as an AgentRuntime backend