Examples¶
Complete, runnable HoloDeck agents live in their own repository:
๐ github.com/justinbarias/holodeck-samples
Each sample ships for OpenAI, Azure OpenAI, and Anthropic (Claude), so you can see the same agent run on either backend.
Available samples¶
| Sample | What it shows |
|---|---|
| Ticket Routing | Structured output, classification, confidence scoring |
| Customer Support | RAG, conversation memory, escalation workflows |
| Content Moderation | Multi-category classification, policy enforcement |
| Legal Summarization | Document analysis, clause extraction, risk identification |
| Legal Assistant | Hierarchical document search, hybrid search, structured citations |
| Financial Assistant | Hierarchical document search, Qdrant native hybrid search, code graders, multi-turn test cases |
Run one¶
git clone https://github.com/justinbarias/holodeck-samples
cd holodeck-samples/financial-assistant/openai # or /claude, /azure
cp .env.sample .env # fill in your credentials
holodeck test run agent.yaml
The OpenAI/Azure samples need the
openai-agentsextra; RAG samples also need a vector-store extra (e.g.qdrant).
Learn the building blocks¶
- Agent Configuration โ the full
agent.yamlstructure - Tools โ function, vectorstore, hierarchical_document, MCP
- Evaluations โ test cases and metrics
- OpenAI Backend ยท Claude Backend