.mint-mb-3 {
    margin-bottom: 0.75rem
}
.mint-mb-4 {
    margin-bottom: 1rem
}
.mint-mt-8 {
    margin-top: 2rem
}
.mint-rounded-lg {
    border-radius: 0.5rem
}
.mint-bg-blue-50 {
    --tw-bg-opacity: 1;
    background-color: rgb(239 246 255 / var(--tw-bg-opacity))
}
.mint-p-6 {
    padding: 1.5rem
}
.mint-font-semibold {
    font-weight: 600
}
.mint-text-blue-600 {
    --tw-text-opacity: 1;
    color: rgb(37 99 235 / var(--tw-text-opacity))
}
.hover\:mint-text-blue-800:hover {
    --tw-text-opacity: 1;
    color: rgb(30 64 175 / var(--tw-text-opacity))
}
.dark\:mint-bg-blue-900\/30:is(.dark *) {
    background-color: rgb(30 58 138 / 0.3)
}
.dark\:mint-text-blue-400:is(.dark *) {
    --tw-text-opacity: 1;
    color: rgb(96 165 250 / var(--tw-text-opacity))
}
.dark\:hover\:mint-text-blue-200:hover:is(.dark *) {
    --tw-text-opacity: 1;
    color: rgb(191 219 254 / var(--tw-text-opacity))
}

Example Configuration

Available Configuration

Generating Text

AgentMark Docs

AgentMark is the git-based Prompt Engineering Platform that empowers both application developers and domain experts to collaborate seamlessly on GenAI products. AgentMark enables companies to manage, evaluate, and improve their full-stack LLM application - with version control, type-safety, and local development built-in.

Overview

Start building awesome GenAI products in under 5 minutes.

Quickstart

The main concepts to help you get started with AgentMark

Core Concepts

Learn how to configure AgentMark for your application needs

Configure custom and non-supported models in AgentMark

Adding Models

Set up a webhook endpoint to test your Agentmark inference in your application

Test Webhook

Learn how AgentMark manages prompts in your application

Learn the core syntax and features of AgentMark

AgentMark Syntax

Learn how to develop and test AgentMark prompts

Development

Learn how to create and use reusable components in AgentMark

Components

Learn how to get structured JSON responses from your prompts

JSON Output

Learn how to extend prompts with tools and create multi-step agents

Tools and Agents

Monitor and debug your prompts with AgentMark

Monitor and debug your prompts using OpenTelemetry

Traces and Logs

Monitor your application with customizable alerts

Alerts

Group related traces together for better observability

Sessions

Analytics provide a high-level overview of how your application is using Generative AI. This includes metrics like costs, tokens, requests, latency, top models, etc.

Metrics

Learn how to evaluate your LLM application using datasets and LLM as a judge evaluations

Datasets enable bulk testing of prompts against diverse inputs and expected outputs

Datasets

Guide to evaluating AI responses with AgentMark

Evaluations

AgentMark allows you to collaborate on prompts while maintaining type safety in a production environment.

Type Safety

Using AgentMark with AI-powered code editors

AI Editors Integration

Run prompts and evaluate results directly in your editor with our VSCode extension

VS Code Extension

Follow the instructions below to install AgentMark in your app.

Getting Started

Learn how AgentMark processes and transforms prompts

Architecture

Generating Object

Generating Image

Generating Speech

AgentMark provides a powerful and flexible way to create prompts using Markdown and JSX. This section will cover the core concepts and features of prompting with AgentMark.

Access and use variables in your prompts using props.

Props

AgentMark supports importing and reusing components across your prompts.

Reusable Components

Conditionals allow you to create dynamic prompts that adapt based on props or other conditions.

Conditionals

AgentMark supports iterating over arrays using the `<ForEach>` tag.

Loops

Define structured output using JSON Schema

Object Schema

Transform values in your prompts using filter functions

Filter Functions

Tools and agents allow you to extend your prompts with external capabilities.

File attachments allow you to attach files to your prompts.

File Attachments

Use Agentmark with Vercel AI SDK to create AI agents with observability.

Vercel AI Integration

Use Agentmark with default adapter to transform prompts into default agentmark format.

Default Integration

Learn how to create your own custom integration with Agentmark.

Custom Integration

Puzzlet is the git-based Prompt Engineering Platform that empowers both application developers and domain experts to collaborate seamlessly on GenAI products. Puzzlet enables companies to manage, evaluate, and improve their full-stack LLM application - with version control, type-safety, and local development built-in.

The main concepts to help you get started with Puzzlet

Learn how to configure Puzzlet for your application needs

Configure models and their user interfaces in Puzzlet

Model Schemas

Set up a webhook endpoint to test your Puzzlet inference in your application

Learn how Puzzlet manages prompts in your application

Learn how to migrate your existing prompts to Puzzlet

Migrating Prompts to Puzzlet

Learn how to develop and test Puzzlet prompts

Learn how to create and use reusable components in Puzzlet

Monitor and debug your prompts with Puzzlet

Guide to evaluating AI responses with Puzzlet

Puzzlet allows you to collaborate on prompts while maintaining type safety in a production environment.

Learn how to use Puzzlet Studio for local development

Puzzlet Studio

Learn how to add model providers to your AgentMark project.

Model Providers

Learn how to migrate your existing LLM application to AgentMark

Migration Guide

Configure model parameters using standard settings.

Model Settings

Execute prompts and get responses from language models

Core API

Observability

AgentMark provides robust type safety through JSON Schema definitions in your prompt files. This ensures type checking for both inputs and outputs, making your prompts more reliable and maintainable.

Tag	Description
`<System>`	System-level instructions
`<User>`	User message
`<Assistant>`	Assistant message

Property	Type	Description	Optional/Required
`model_name`	`string`	The name of the model to use for text generation.	Required
`max_tokens`	`number`	Maximum number of tokens to generate.	Optional
`temperature`	`number`	Controls the randomness of the output; higher values result in more random outputs.	Optional
`max_calls`	`number`	Maximum number of LLM calls allowed.	Optional
`top_p`	`number`	Controls the cumulative probability for nucleus sampling.	Optional
`top_k`	`number`	Limits the next token selection to the top `k` tokens.	Optional
`presence_penalty`	`number`	Penalizes new tokens based on their presence in the text so far, encouraging the model to discuss new topics.	Optional
`frequency_penalty`	`number`	Penalizes new tokens based on their frequency in the text so far, reducing the likelihood of repeating the same line verbatim.	Optional
`stop_sequences`	`string[]`	Array of strings where the generation will stop if any of the strings are encountered.	Optional
`seed`	`number`	Seed value for random number generation, ensuring reproducibility.	Optional
`max_retries`	`number`	Maximum number of retries for the request in case of failures.	Optional
`tool_choice`	`"auto" \| "none" \| "required" \| { type: "tool", tool_name: string }`	Controls how tools are used in the generation process.	Optional
`tools`	`Record<string, { description: string, parameters: JSONSchema }>`	Defines available tools and their parameters for the model to use.	Optional

Getting Started

Generation Types

Prompt Basics

Running Prompts

Generating Text

Example Configuration

Tags

Available Configuration

Have Questions?

Getting Started

Generation Types

Prompt Basics

Running Prompts

​Example Configuration

​Tags

​Available Configuration

Have Questions?

Example Configuration

Tags

Available Configuration