announcement,

Shipping PPL: A License for the Age of PromptWar̊e ØS

Huan Li Huan Li Follow Dec 13, 2025 · 3 mins read
Shipping PPL: A License for the Age of PromptWar̊e ØS

Where is the GPL for Prompts?

Last week, I wrote From GPL to Prompts to highlight a growing problem: English is becoming source code, but our open source licenses are stuck in the era of C and Java.

I asked the question. Today, I am sharing the answer.

We are releasing the Public Prompt License (PPL), a new family of open-source licenses designed specifically for the AI-Native era.


The PromptWar̊e ØS Trigger

As we were building PromptWar̊e ØS—our operating system for AI co-founders—we hit a wall. We wanted to share our “kernel” prompts and “skills” libraries as true open source.

But when we looked at the existing options, they didn’t fit:

  • MIT is great, but it doesn’t define what “source” means for an agent.
  • AGPL is powerful, but ambiguous. If you talk to my agent over an API, do I have to send you the Python wrapper or the System Prompt?
  • SSPL pioneered solving the cloud gap with bold scope—claiming the entire “Service Stack” to ensure complete freedom. While we deeply respect this approach, its rejection by the Open Source Initiative (OSI) limits adoption in some ecosystems we want to serve.

We needed something that protected the soul of the agent (the prompts) while being pragmatic about OSI approval and ecosystem adoption.

Enter PPL v0.1

So we drafted the Public Prompt License (PPL).

It is not just a text file; it is a “Legal Spec” for AI agents.

The Core Innovation: “Prompt Source”

The heart of PPL is a rigorous definition of Prompt Source. It explicitly includes:

  • System Prompts & Personas
  • Tool Descriptions & Schemas
  • Routing Logic & Orchestration
  • Knowledge Structure (like PromptWar̊e ØS’s bootloaders)

Crucially, it explicitly excludes the underlying inference engine (vLLM, Ollama), the OS, and the hardware.

This is our Infrastructure Exclusion. SSPL took the maximalist approach—ensuring complete freedom by requiring disclosure of the entire service stack. PPL takes a surgical approach: we protect the cognitive logic (the prompts) while being pragmatic about OSI approval. Both philosophies have merit; we chose the path that balances idealism with ecosystem adoption.

The Three Variants

We created a family of three licenses to match the software ecosystem:

  1. PPL-M (MIT-style): For the builders. Use these prompts anywhere, even in closed products. Just keep the credit.
  2. PPL-A (Apache-style): For the enterprise. Includes patent grants for prompting techniques and clear attribution.
  3. PPL-S (Service-style): For the commons. If you run this agent as a service, you must share the prompts. This ensures that the knowledge embedded in our agents remains open.

License for English

Why Not SSPL?

We considered SSPL carefully. MongoDB’s Server Side Public License embodies an important open-source principle—preventing cloud giants from freeloading on the commons without contributing back.

But we chose a different path for two reasons:

  1. OSI Approval: We want PPL to be recognized as OSI-approved open source, which means working within their definition, even if we don’t entirely agree with all aspects of it.
  2. Scope Precision: For AI agents, we need a definition that’s surgically precise about what “Prompt Source” means. SSPL’s broad “Service Stack” definition works for databases; we needed something tailored for cognitive systems.

This doesn’t mean SSPL is wrong. It means we’re optimizing for different constraints.

An Invitation to Pilot

This is not just legal theory. We are applying PPL to PromptWar̊e ØS (PromptWare OS) and our internal Ship.Fail projects starting today.

We are releasing PPL as Draft 0.1 (Request for Comment). We know it isn’t perfect yet. That is why we need you.

We invite you to:

  1. Read the Spec: Visit the project page or check out the DEFINITIONS.md in the repo.
  2. Pilot it: If you are building an agent, try adding PPL-M or PPL-S to your repo.
  3. Break it: Tell us where the definitions fail.

The future of open source isn’t just about code anymore. It’s about the prompts that steer our world.

Let’s build that commons together.

👉 Visit the Project Page Star on GitHub
Join Newsletter
Get the latest news right in your inbox. We never spam!
Written by Huan Li Follow
Founding Human & Architect | Author @ Wechaty, Silicon Adventist, Serial Entrepreneur, Angel Investor, Burner🔥