Welcome to the PRPM Wiki
Relevant source files
The following files were used as context for generating this wiki page:
- DOCUMENTATION_MAP.md
- packages/webapp/src/app/blog/agents-md-deep-dive/page.tsx
- packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx
- packages/webapp/src/app/blog/windsurf-deep-dive/page.tsx
- packages/webapp/src/app/blog/copilot-deep-dive/page.tsx
- packages/webapp/src/app/blog/format-specifications-guide/page.tsx
- CONTRIBUTING.md
- packages/webapp/src/app/blog/write-once-distribute-everywhere/page.tsx
- public-documentation/mint.json
- packages/webapp/src/app/blog/distributable-intelligence/page.tsx
- packages/webapp/src/app/blog/kiro-deep-dive/page.tsx
- packages/webapp/src/app/blog/introducing-prpm/page.tsx
- packages/webapp/src/app/blog/continue-deep-dive/page.tsx
- packages/webapp/public/blog/prpm-complete-feature-guide.md
- packages/webapp/src/app/blog/discovering-windsurf-rules/page.tsx
- packages/webapp/src/app/blog/openskills-cross-ide-compatibility/page.tsx
- packages/webapp/public/legal/ACCEPTABLE_USE_POLICY.md
PRPM (Prompt Package Manager) is a universal registry and package manager for AI prompts, agents, skills, and rules. Its primary goal is to solve the fragmentation problem in the AI coding assistant ecosystem, where each tool (like Cursor, Claude, Continue, etc.) uses its own proprietary format for instructions. PRPM provides a centralized platform to discover, share, and manage these AI assets, making them work across different editors and tools.
Sources: packages/webapp/src/app/blog/introducing-prpm/page.tsx:28-40
The core philosophy of PRPM is "Write Once, Distribute Everywhere." It allows authors to publish an AI coding package in a single format, and the PRPM system handles the conversion to all other supported formats on-the-fly. This enables developers to use consistent AI guidance across a team, even if they use different AI coding assistants, and it significantly reduces the maintenance burden for package authors.
Sources: packages/webapp/src/app/blog/write-once-distribute-everywhere/page.tsx:43-47, packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:43-47
What is PRPM?
PRPM is a platform designed to manage "distributable intelligence" for AI-assisted development. As AI assistants become capable of refactoring entire codebases, they lack the framework-specific and company-specific patterns to do so correctly. PRPM bridges this gap by allowing maintainers to publish executable knowledge in the form of versioned packages.
Sources: packages/webapp/src/app/blog/distributable-intelligence/page.tsx:32-42
These packages can contain:
- Rules: Declarative constraints that AI enforces during code generation.
- Skills: Step-by-step procedures for specific, repeatable tasks.
- Agents: Complex, multi-step orchestration logic with reporting and edge-case detection.
Sources: packages/webapp/src/app/blog/distributable-intelligence/page.tsx:143-189
The platform consists of a CLI, a web application with a package registry, and a testing playground.
Sources: packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:29-32
Key Features
| Feature | Description | Source |
|---|---|---|
| Universal Package Management | Install a package once, and PRPM automatically converts it to the format required by your editor (Cursor, Claude, Continue, etc.). | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:49-62 |
| Format Conversion | A powerful engine that performs semantically-aware, lossless conversions between all 8 major AI IDE formats. | packages/webapp/src/app/blog/write-once-distribute-everywhere/page.tsx:53-57 |
| AI-Powered Search | A hybrid search system combining traditional keyword search with AI semantic search to understand user intent and find relevant packages. | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:87-99 |
| The Playground | A web and CLI tool to test packages with real AI models (GPT-4, Claude Sonnet, etc.) before installation. | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:127-138 |
| Collections | Bundles of related packages that can be installed with a single command to set up a complete workflow for a specific stack (e.g., Next.js). | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:109-115 |
| Quality Scoring | An automated system that rates packages (0-5 stars) based on content depth, author credibility, community engagement, and maintenance. | packages/webapp/public/blog/prpm-complete-feature-guide.md:65-83 |
| Author Analytics | Verified authors get access to detailed analytics on downloads, views, stars, and playground usage. | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:179-182 |
| Organizations & Private Packages | Teams can create organizations to manage and publish private packages accessible only to their members. | packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:201-204 |
Core Concepts
Packages
A package is the fundamental unit in PRPM. It contains AI instructions, metadata, and content. Packages are defined by a package.prpm.json manifest.
Sources: CONTRIBUTING.md:13-27
{
"name": "my-awesome-rule",
"version": "1.0.0",
"description": "A helpful Cursor rule for...",
"type": "cursor",
"category": "code-quality",
"tags": ["typescript", "testing", "best-practices"],
"author": "your-username",
"license": "MIT",
"content": "./rule.md"
}
Source: CONTRIBUTING.md:16-26
Format Conversion
The "Write Once, Distribute Everywhere" workflow is powered by PRPM's format conversion engine. It understands the nuances of each AI IDE's format and translates between them.
This diagram illustrates the conversion flow when a user installs a package.
Supported Formats
PRPM supports conversion between 8 major AI IDE prompt formats.
| Format | File Location(s) | Frontmatter | Key Features |
|---|---|---|---|
| Cursor | .cursor/rules | Required | 4 rule types, MDC format |
| Claude Code | .claude/{agents,skills,commands}/ | Required | allowed-tools, model selection, hooks |
| Continue | .continue/rules/*.md | Required | Globs, regex matching, alwaysApply logic |
| Windsurf | .windsurf/rules | None | Plain markdown, 12k character limit |
| GitHub Copilot | .github/copilot-instructions.md | Optional | Two-tier system, path-specific rules |
| Kiro Steering | .kiro/steering/*.md | Optional | Inclusion modes, foundational files (product.md, tech.md) |
| Kiro Hooks | .kiro/hooks/*.json | N/A (JSON) | Event-driven file automations (e.g., on file creation) |
| agents.md | agents.md | None | Open standard, plain markdown, tool-agnostic |
| Source: packages/webapp/src/app/blog/format-specifications-guide/page.tsx:101-170 |
Getting Started
To start using PRPM, install the CLI via npm.
# Install the CLI globally
npm install -g prpm
# Search for packages
prpm search "react best practices"
# Install a package (it will auto-convert to your editor's format)
prpm install @vendor/package-name
# Test a package in the playground before installing
prpm playground --package @vendor/package-name
Source: packages/webapp/src/app/blog/prpm-complete-feature-guide/page.tsx:215-231
The following sequence diagram shows the user flow for installing a package.
Documentation Structure
The project's documentation is split into user-facing and internal development docs. This structure helps separate information based on the audience's role.
This diagram shows the high-level documentation layout.
Source: DOCUMENTATION_MAP.md:21-62
Quick Links by Role
- Users: Start with
README.md, then move todocs/INSTALLATION.mdanddocs/CLI.md. - Package Authors: Refer to
docs/PUBLISHING.mdanddocs/COLLECTIONS.md. - Contributors: Read
CONTRIBUTING.mdanddevelopment/docs/DEVELOPMENT.md. - Deployers: Use the guides in
development/docs/, starting withDEPLOYMENT_SUMMARY.md.
Source: DOCUMENTATION_MAP.md:68-91
Conclusion
PRPM aims to be the essential infrastructure for sharing and managing "distributable intelligence" in the age of AI-assisted software development. By providing a universal registry, robust tooling, and a seamless cross-platform experience, it empowers developers to codify and share best practices, making AI assistants more powerful and context-aware. The platform solves the critical problem of format fragmentation, benefiting package authors, users, and the entire AI coding community.
Sources: packages/webapp/src/app/blog/distributable-intelligence/page.tsx:32-42, 260-262, packages/webapp/src/app/blog/introducing-prpm/page.tsx:34-47