Examples Overview
May 7, 2026 · View on GitHub
This page lists all maintained Hayhooks examples with detailed descriptions and links to the source code.
Pipeline wrapper examples
| Example | Docs | Code | Description |
|---|---|---|---|
| Chat with Website (Streaming) | chat-with-website.md | GitHub | Website Q&A with streaming |
| Chat with Website (basic) | — | GitHub | Minimal website Q&A wrapper |
| Chat with Website (MCP) | — | GitHub | Exposes website Q&A as MCP Tool |
| Async Question Answer (Streaming) | async-operations.md | GitHub | Async pipeline and streaming patterns |
| Custom Tracing | tracing.md | GitHub | Direct OpenAI-based YAML classifier (io_bound vs cpu_bound) with nested custom tracing spans |
| Open WebUI Agent Events | openwebui-events.md | GitHub | UI events and status updates |
| Open WebUI Agent on Tool Calls | openwebui-events.md | GitHub | Tool call interception & feedback |
| Open WebUI Reasoning Agent (GPT-5.4 mini) | openai-compatibility.md | GitHub | Streams reasoning summaries to Open WebUI "Thinking" blocks using OpenAIResponsesChatGenerator |
| Responses File Upload (Server-side tools) | openai-compatibility.md | GitHub | Responses API file reading/upload via server-side agent tools |
| Agent Codex (Client-Side Tools) | openai-compatibility.md | GitHub | Hybrid Responses API demo: server weather tool + Codex client-side tools (exec_command) |
| Image Generation (File Response) | file-response-support.md | GitHub | Returning binary files (images) from run_api |
| Shared Code Between Wrappers | — | GitHub | Reusing code across wrappers |
End-to-end examples & patterns
| Example | Docs | Code | Description |
|---|---|---|---|
| RAG: Indexing and Query with Elasticsearch | rag-system.md | GitHub | Full indexing/query pipelines with Elasticsearch |
| API Key Authentication | advanced-configuration.md | GitHub | Middleware-based API key auth with multi-key support and Swagger Authorize |
How to use examples
Prerequisites:
- Install Hayhooks:
pip install hayhooks(additional deps per example may apply)
Deployment:
- Pipeline wrappers: deploy directly with
hayhooks pipeline deploy-files -n <name> <example_dir>and run via API (POST /<name>/run) or OpenAI-compatible chat endpoints if implemented - End-to-end examples: follow the example's documentation for full setup (services like Elasticsearch, multi-pipeline deployment, datasets, etc.)
For general usage and CLI commands, see the Getting Started Guide.
Try all examples with Docker Compose
The repository includes a compose.yml file that automatically deploys all pipeline wrapper examples:
# Clone the repository
git clone https://github.com/deepset-ai/hayhooks.git
cd hayhooks
# Set your API key (required for most examples)
export OPENAI_API_KEY=your-api-key
# Start Hayhooks with all examples
docker compose up -d
This will:
- Start Hayhooks on port
1416(API) and1417(MCP server) - Auto-deploy all pipeline wrappers from
examples/pipeline_wrappers/ - Install required dependencies like
trafilatura
Check deployed pipelines:
curl localhost:1416/status
Access the API documentation at http://localhost:1416/docs.