This directory contains small scripts that demonstrate different agent patterns. Run them with pnpm using the commands shown below.

• agents-as-tools.ts – Orchestrate translator agents using them as tools.

  pnpm examples:agents-as-tools

• agents-as-tools-structured.ts – Use structured tool input with Agent.asTool().

  pnpm examples:agents-as-tools-structured

• agents-as-tools-conditional.ts – Enable language tools based on user preference.

  pnpm examples:agents-as-tools-conditional

• deterministic.ts – Fixed agent flow with gating and quality checks.

  pnpm examples:deterministic

• forcing-tool-use.ts – Require specific tools before final output.

  pnpm -F agent-patterns start:forcing-tool-use

• human-in-the-loop.ts – Manually approve certain tool calls.

  pnpm examples:human-in-the-loop

• human-in-the-loop-stream.ts – Streaming version of human approval.

  pnpm examples:streamed:human-in-the-loop

• input-guardrails.ts – Reject unwanted requests with guardrails.

  pnpm examples:input-guardrails

• llm-as-a-judge.ts – Evaluate and iterate on story outlines.

  pnpm -F agent-patterns start:llm-as-a-judge

• output-guardrails.ts – Block unsafe output using guardrails.

  pnpm examples:output-guardrails

• parallelization.ts – Run translations in parallel and pick the best.

  pnpm examples:parallelization

• routing.ts – Route messages to language-specific agents.

  pnpm examples:routing

• streamed.ts – Stream agent output, both text and events.

  pnpm examples:streamed

• streaming-guardrails.ts – Check streaming output against guardrails.

  pnpm -F agent-patterns start:streaming-guardrails