Skip to main content

Tactus

Examples / Agents

Agents

This chapter focuses on working with AI agents in Tactus - the core abstraction for interacting with language models.

Examples in This Chapter

Multi Model

Demonstrates using multiple models in a single workflow - a researcher using GPT-4o and a summarizer using GPT-4o-mini. This example shows:

  • Defining multiple agents with different models
  • Sequential agent execution (research then summarize)
  • Template strings in initial messages ({input.topic}, {research})
  • Passing results between agents
  • Turn-limited loops for each agent phase
  • Using the done tool to extract agent outputs

Has Specs · Requires API Keys

Streaming

Shows real-time streaming responses from an agent. This example demonstrates:

  • Enabling streaming mode with streaming = true
  • Processing token-by-token responses as they arrive
  • Accumulating streamed content into a final result
  • Using streaming for interactive or long-form generation

Has Specs · Requires API Keys

Bedrock

Demonstrates using AWS Bedrock as a provider with multiple Claude models. This example shows:

  • Configuring the bedrock provider
  • Using different Claude models (Claude 3.5 Sonnet, Claude 3 Haiku, Claude 3 Opus)
  • AWS credential configuration requirements
  • Provider-specific model identifiers
  • Multi-agent workflows across different Bedrock models

Has Specs · Requires API Keys

Want to contribute?

The examples repository is open source. Add your own examples or improve existing ones.

View on GitHub