Skip to content
ALTAIRA LABS
DocsBlog
PromptPackPromptPackPromptPack

PromptPack Ecosystem

PromptPack is a specification, not a product or framework. The ecosystem consists of the spec itself and tools that work with it.

Core Components

PromptPack Specification

Status: v1.4 Available Repository: altairalabs/promptpack-spec

The open specification defining the JSON format for packaging AI agent behavior — from multi-prompt routers to autonomous agent loops. Includes:

  • Complete schema definition
  • RFC process for evolution
  • Documentation and examples
  • Governance model

View Specification →

PromptKit Toolkit

Status: Available Now Separate Project: Related but independent

The reference toolkit for loading, executing, testing, and compiling PromptPacks. Ships:

  • promptarena CLI — multi-provider testing, red-team scenarios, CI/CD reports
  • packc CLI — compile workspace config into portable .pack.json artifacts
  • Go runtime for executing packs with workflows, agent loops, tools, and skills
  • GitHub Actions for CI/CD integration
Terminal window
npm install -g @altairalabs/promptarena @altairalabs/packc

Learn more →

PromptArena Testing

Status: Available Now Part of: PromptKit toolkit

CLI tool for systematic testing of agent behavior across multiple providers:

  • Multi-provider comparison (OpenAI, Anthropic, Google, etc.)
  • Multi-turn conversation testing
  • Multimodal content testing (images, audio, video)
  • Self-play mode for realistic simulations
  • Comprehensive reporting (HTML, JSON, JUnit, Markdown)
  • CI/CD integration

Get Started with PromptArena →

Relationship Between Components

flowchart TB
    spec["<b>PromptPack Specification</b><br/>JSON format · schema · validation · docs"]
    runtime["<b>PromptKit Runtime</b><br/>executes packs · SDK libraries · validation tools"]
    arena["<b>PromptArena Testing</b><br/>multi-provider testing · simulation · reports"]


    spec -- implements &amp; tests --> runtime
    runtime -- uses for testing --> arena

These are separate but related projects:

  • PromptPack Spec: Defines the standard (open governance, RFC process)
  • PromptKit: Reference toolkit — shipped and actively maintained by AltairaLabs
  • PromptArena: Testing CLI shipped as part of the PromptKit toolkit

Community Tools

As PromptPack adoption grows, the community is building compatible tools and integrations.

Official Integrations

LangChain.js Integration

Status: Available Now Package: @promptpack/langchain Repository: altairalabs/promptpack-langchainjs

Native LangChain.js integration for loading and using PromptPacks with LangChain. Features:

  • Prompt Registry - Load and manage PromptPacks from JSON/YAML
  • Tool Integration - Built-in tool calling with governance policies
  • Validation - Response validation and guardrails
  • LangChain Native - Integrates with LangChain templates
  • Type Safe - Full TypeScript support

Installation:

Terminal window
npm install @promptpack/langchain

Quick Example:

import { PromptPackRegistry, PromptPackTemplate } from '@promptpack/langchain';
import { ChatOpenAI } from '@langchain/openai';


// Load the pack
const pack = PromptPackRegistry.loadFromFile('./customer-support.json');


// Create a template
const template = new PromptPackTemplate({
 pack,
 promptId: 'support',
});


// Use with LangChain
const model = new ChatOpenAI({ modelName: 'gpt-4' });
const chain = template.pipe(model);


// Invoke with variables
const response = await chain.invoke({
 role: 'helpful support agent',
 company: 'TechCorp',
 issue_type: 'billing',
});

View Documentation →

Additional Community Tools

As the ecosystem grows, we expect more community-contributed tools including:

  • Alternative runtimes in different languages
  • Framework integrations (LlamaIndex, Transformers)
  • IDE plugins and extensions
  • Deployment tools and platforms
  • Monitoring and observability solutions

Want to build PromptPack tooling? The specification is open—anyone can build compatible tools. See our Contributing Guide to get involved.

Planned Integrations

LLM Providers

  • OpenAI (GPT-4, GPT-3.5)
  • Anthropic (Claude)
  • Google (Gemini)
  • Azure OpenAI
  • Local models (Ollama, LM Studio)

Orchestration Frameworks

  • LangChain.js adapter (available now - see above)
  • LlamaIndex adapter (proposal in backlog)
  • Transformers adapter (proposal in backlog)
  • Custom framework integrations

Development Tools

  • VS Code Extension
  • CLI validation and linting tools
  • Testing frameworks
  • CI/CD pipeline integrations

Integration Philosophy

PromptPack is a file format specification, not an implementation. It defines how to structure prompts, tools, and workflows in a standardized JSON format. The specification does NOT include LLM provider SDKs, API integrations, or hosted services.

Any tool can implement PromptPack support by:

  1. Reading PromptPack JSON files
  2. Validating against the JSON Schema
  3. Building their own runtime/execution engine
  4. Implementing their own provider integrations

Future Plans

The PromptPack ecosystem roadmap includes:

  • Growing the compatible tools ecosystem
  • Additional framework adapters (LlamaIndex, Transformers)
  • Pack registry / PromptPack Hub for community sharing
  • Continued specification evolution via RFC process

We’re in the early stages of building an ecosystem. Join us in shaping it.

View Roadmap →

Contributing

Interested in building an integration?

  1. Check the Specification - Ensure you understand the format
  2. Build Your Integration - As a separate project/package
  3. Share with the Community - Via discussions or PRs to add documentation
  4. Maintain Compatibility - Follow semantic versioning of the spec

See our Contributing Guide to get involved.