Skills lab

Demos

• 01

  Specification precision

  System prompt library and Zod-validated structured output with generateText and Output.object.

  Open demo

• 02

  Prompt engineering patterns

  Compare zero-shot, few-shot, and chain-of-thought prompt patterns side-by-side.

  Open demo

• 03

  Embeddings & vector search

  Semantic search API with embed / embedMany and a vector store.

  Open demo

• 04

  RAG pipeline

  End-to-end retrieve-augment-generate with streaming responses.

• 05

  Evaluation

  Eval harness with LLM-as-a-judge and retrieval metrics.

• 06

  Single agent

  ReAct-style agent with tools, steps, and memory.

  Open demo

• 07

  Multi-agent

  Planner–worker orchestration with verification checkpoints.

  Open demo

• 08

  Failure patterns

  Demos and fixes for common agent failure modes.

  Open demo

• 09

  Context architecture

  Multi-source RAG with metadata filtering.

• 10

  Guardrails

  Input/output validation and human-in-the-loop approval flows.

  Open demo

• 11

  Observability

  Tracing, latency, tokens, and cost visibility.

  Open demo

• 12

  Token economics

  Model cost comparison and routing by task complexity.

  Open demo

• 13

  Fine-tuning

  Small fine-tune vs RAG comparison on a focused task.

• 14

  MCP

  HTTP MCP server plus an agent client that discovers Chuck Norris API tools at runtime.

  Open demo

• 15

  Deploy & ship

  Live demos, repo narrative, and portfolio polish.