Quickstart

Welcome to PromptTrail! Choose your path:

Quick Navigation

1. Just want to call LLM APIs?

   • Start with Preparation to set up your environment

   • Then read Models to learn how to make API calls

   • For production:

     • Learn about Caching to optimize API usage

     • Explore Tool/Function Calling for advanced features

2. Want to build complex conversational agents?

   • Follow path 1 first to understand the basics

   • Then dive into Agents for building conversational flows

Table of Contents

• Preparation
  • Install
• models: Unified Interface to LLM API
  • Make an API Call
  • Core Concepts
    • Message
    • Session
    • Model and Configuration
  • Try different API
    • Google
    • Anthropic
  • Try local LLMs
  • Stream Output
• agents: Agent as Code
  • Introduction
    • Template
    • Runners
      • CommandLineRunner
      • APIRunner
      • Common Runner Features
  • Subroutines
    • Session Initialization Strategies
    • Message Squashing Strategies
    • Using SubroutineTemplate
  • Hooks
  • Rendering
  • Session
  • Control Flow
  • Tool (Function Calling)
    • Subroutine Tool
• core: Developer Tools for LLMs
  • Metadata
  • Cache
  • Mock
  • Debug and Logging
    • Debug Mode
    • Logging
    • Debuggable Base Class

Project Structure

PromptTrail is organized into several core modules:

• prompttrail.models: Unified interfaces to various LLM providers

• prompttrail.agent: Framework for building complex LLM applications

• prompttrail.core: Core utilities and developer tools

Each module has its own quickstart guide above.