feat: Add experimental AG-UI supported chat endpoint and PPL query assist

Dockerfile

@@ -148,6 +148,7 @@ RUN git config --global core.symlinks false
 RUN rm -rf /usr/local/lib/python3.11/site-packages/setuptools-65.5.1.dist-info
 RUN rm -rf /usr/local/lib/python3.11/ensurepip/_bundled/setuptools-65.5.0-py3-none-any.whl
 
+COPY ./experimental/ag-ui/server-agui.py /app/experimental/ag-ui/server-agui.py
 COPY ./holmes /app/holmes
 COPY ./server.py /app/server.py
 COPY ./holmes_cli.py /app/holmes_cli.py

experimental/ag-ui/README.md

@@ -0,0 +1,75 @@
+# AG-UI - Experimental HolmesGPT Server
+
+AG-UI is an experimental extension to HolmesGPT that demonstrates [AG-UI](https://docs.ag-ui.com/introduction) capabilities (page context and state sharing, front-end tools execution, etc) through a specialized `/api/agui/chat` endpoint and a web-based "ExampleOps✨" observability demo interface. The AG-UI compatible `server.py` is adapted from the [existing server.py implementation](../../server.py)
+
+_⚠️ **Disclaimer**: AG-UI is experimental within HolmesGPT. APIs and interfaces may change as the project evolves. The demonstration server and frontend is not intended for production use._
+
+![ExampleOps demo video](https://github.com/kylehounslow/holmesgpt/blob/docs/experimental/ag-ui/docs/holmesgpt-agui-demo-1.gif?raw=true)
+
+## Known Limitations:
+
+1. Front-end tool auto-discovery and integration with LLM is not yet supported. Front-end tool discovery and execution is handled statically in the back-end.
+1. Front-end tool results are not yet supported. No-op 200 response is returned to front-end.
+1. Tested with Anthropic Claude Sonnet 3.7 and 4.0 on AWS Bedrock only.
+1. `TodoWrite` tool rendering is not properly supported.
+
+## 🛠️ Quick Start
+
+### **0. Prerequisites**
+
+- **HolmesGPT** with experimental AG-UI server.
+- **Data Sources**: Prometheus (`:9090`) and/or OpenSearch (`:9200`)
+  - Recommended to run [opentelemetry-demo](https://github.com/open-telemetry/opentelemetry-demo) via docker-compose.
+- **Node.js** 20+ (for frontend demonstration)
+
+### **1. Set up datasources**
+
+A simple way to generate synthetic Prometheus metrics, OpenSearch logs, traces and more, is to use: [opentelemetry-demo](https://github.com/open-telemetry/opentelemetry-demo)
+
+```bash
+git clone git@github.com:open-telemetry/opentelemetry-demo.git
+cd opentelemetry-demo
+docker compose up -d
+```
+
+### **2. Start HolmesGPT AG-UI Server on port 5050**
+
+Assumes there is a local Prometheus server (containing metrics) at `localhost:9090` (e.g. [opentelemetry-demo](https://github.com/open-telemetry/opentelemetry-demo)) and using AWS Bedrock as LLM. See [HolmesGPT documentation](https://holmesgpt.dev/) for alternate configurations.
+
+```bash
+# Start HolmesGPT AG-UI compatible server
+cd holmesgpt
+export HOLMES_PORT=5050
+export PROMETHEUS_URL=http://localhost:9090
+export AWS_PROFILE=default
+export AWS_REGION=us-east-1
+export MODEL=bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0
+poetry run python experimental/ag-ui/server-agui.py
+```
+
+### **3. Run "ExampleOps✨" Demo Frontend**
+
+ExampleOps✨ is a lightweight observability frontend that demonstrates AG-UI capabilities (page context, state sharing, front-end tools execution, etc).
+Create .env file at `experimental/ag-ui/front-end/.env`. See example below and replace Prometheus/OpenSearch urls as needed:
+
+```env
+# AG-UI Agent Configuration
+HOLMES_PORT=5050
+AGENT_URL=http://localhost:${HOLMES_PORT}
+# Prometheus Configuration
+REACT_APP_PROMETHEUS_URL=http://localhost:9090
+
+# OpenSearch Configuration
+REACT_APP_OPENSEARCH_URL=http://localhost:9200
+REACT_APP_OPENSEARCH_USER=user
+REACT_APP_OPENSEARCH_PASSWORD=pass
+```
+
+```bash
+cd experimental/ag-ui/front-end
+npm install && npm start
+```
+
+## Running HolmesGPT in OpenSearch Dashboards
+OpenSearch Dashboards (>=3.3) supports AG-UI compatible agents for its AI chat. For an example HolmesGPT configuration,
+see [gist](https://gist.github.com/kylehounslow/07290ee15768a5b15a924831f7759217).

experimental/ag-ui/front-end/.env.example

@@ -0,0 +1,12 @@
+# AG-UI Front-End Configuration
+PORT=5151
+# AG-UI Agent Configuration
+HOLMES_PORT=5050
+AGENT_URL=http://localhost:${HOLMES_PORT}
+# Prometheus Configuration
+REACT_APP_PROMETHEUS_URL=http://localhost:9090
+
+# OpenSearch Configuration
+REACT_APP_OPENSEARCH_URL=http://localhost:9200
+REACT_APP_OPENSEARCH_USER=admin
+REACT_APP_OPENSEARCH_PASSWORD=admin

experimental/ag-ui/front-end/.gitignore

@@ -0,0 +1,3 @@
+node_modules/
+package-lock.json
+.env

experimental/ag-ui/front-end/.nvmrc

@@ -0,0 +1 @@
+20

experimental/ag-ui/front-end/README.md

@@ -0,0 +1,42 @@
+# ExampleOps✨
+
+A React TypeScript demo observability application with integrated HolmesGPT AG-UI chat assistant.
+
+![ExampleOps demo video](https://github.com/kylehounslow/holmesgpt/blob/docs/experimental/ag-ui/docs/holmesgpt-agui-demo-1.gif?raw=true)
+
+## Setup
+
+1. Install dependencies:
+   ```bash
+   npm install
+   ```
+
+2. Configure your agent endpoint:
+   ```bash
+   cp .env.example .env
+   # Edit .env to set your AGENT_URL
+   ```
+
+3. Start the development server:
+   ```bash
+   npm start
+   ```
+
+## Configuration
+
+The app expects an AG-UI compatible agent service running at the base URL specified in `AGENT_URL`. The chat endpoint will be accessed at `${AGENT_URL}/api/agui/chat`. The agent should:
+
+- Follow the AG-UI Protocol
+- Accept POST requests with `RunAgentInput` payload
+- Return Server-Sent Events (SSE) stream containing AG-UI events
+
+
+## AG-UI Integration
+
+The chat assistant uses:
+
+- **HttpAgent**: For HTTP-based agent communication
+- **AgentSubscriber**: For handling streaming events
+- **Event Types**: TEXT_MESSAGE_START, TEXT_MESSAGE_CONTENT, TEXT_MESSAGE_END, etc.
+
+See the [AG-UI documentation](https://docs.ag-ui.com) for more details.

experimental/ag-ui/front-end/package.json

@@ -0,0 +1,45 @@
+{
+  "name": "ag-ui-app",
+  "version": "0.1.0",
+  "private": true,
+  "dependencies": {
+    "@ag-ui/client": "^0.0.38",
+    "@ag-ui/core": "^0.0.38",
+    "@types/node": "^16.18.126",
+    "@types/react": "^18.3.24",
+    "@types/react-dom": "^18.3.7",
+    "chart.js": "^4.5.0",
+    "chartjs-adapter-date-fns": "^3.0.0",
+    "date-fns": "^2.30.0",
+    "react": "^18.3.1",
+    "react-chartjs-2": "^5.3.0",
+    "react-dom": "^18.3.1",
+    "react-markdown": "^9.0.1",
+    "react-scripts": "^5.0.1",
+    "typescript": "^4.9.5"
+  },
+  "scripts": {
+    "start": "react-scripts start",
+    "build": "react-scripts build",
+    "test": "react-scripts test",
+    "eject": "react-scripts eject"
+  },
+  "eslintConfig": {
+    "extends": [
+      "react-app",
+      "react-app/jest"
+    ]
+  },
+  "browserslist": {
+    "production": [
+      ">0.2%",
+      "not dead",
+      "not op_mini all"
+    ],
+    "development": [
+      "last 1 chrome version",
+      "last 1 firefox version",
+      "last 1 safari version"
+    ]
+  }
+}

experimental/ag-ui/front-end/public/index.html

@@ -0,0 +1,12 @@
+<!DOCTYPE html>
+<html lang="en">
+  <head>
+    <meta charset="utf-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1" />
+    <title>ExampleOps✨</title>
+  </head>
+  <body>
+    <noscript>You need to enable JavaScript to run this app.</noscript>
+    <div id="root"></div>
+  </body>
+</html>