The web didn’t die. It was smothered.
We’re taking the air back.
🚀 Get Started — Build your first semantic artefact
⚡ Read the Manifesto — Why we're building this
Every day, more of the open web is being swallowed by walled gardens, algorithmic sludge, and AI engines that strip-mine meaning but give nothing back.
Creators lose traffic. Publishers lose revenue. Readers lose context.
And the web becomes a shallow pool of recycled noise — increasingly unreadable, unfollowable, and untrustworthy.
SPP is a new publishing standard for a very different future.
It lets content be discovered, reconstructed, trusted, and understood — by both humans and machines — without sacrificing ownership, context, or provenance.
SPP is a decentralised, AI-native content protocol that turns passive web pages into active, intelligent, and self-defending artefacts. It makes your content:
- 🧠 Machine-readable — ready to be reconstructed and reused by any AI, without guessing
- 🔐 Owner-verifiable — no more scraping, no more theft; attribution is built in
- 🧭 Context-aware — agents know where, when, and why something was published
- 🤝 Endorseable & trustable — with signals from real humans, not black-box algorithms
It’s RSS with a brain. JSON-LD with attitude. ActivityPub that actually works.
SPP isn’t just a spec. It’s a rebellion in structured form.
For Creators & Publishers:
Your content becomes a live, verifiable data stream. AI agents can subscribe, remix it (with credit), and even pay you — no more platform middlemen. You own the whole journey.
For Developers & AI Engineers:
Stop scraping brittle HTML. Stop reverse-engineering meaning. With SPP, you get clean, structured truth you can build on — whether you're feeding LLMs, powering dashboards, or creating agentic assistants.
For Everyone Else:
This is the way back to a web that works for us. A web where trust is built in, credit flows to the source, and discovery is driven by meaning, not manipulation.
Stop begging platforms for scraps. Start owning the conversation.
Instead of publishing a messy HTML page and hoping an AI interprets it right, you publish this:
{
"claim": "The current interest rate is 5.25%",
"source": "entity:federal-reserve-announcement-123",
"published_at": "2025-08-07T14:00:00Z"
}Now any AI agent can subscribe to your feed and know, with certainty and provenance, what the interest rate is.
No scraping. No guessing. Just truth.
We’re at a fork in the road.
The open web is collapsing into a graveyard of dead links, ghost-town blogs, and SEO-choked clickbait.
Platforms throttle external links. AI models regurgitate without credit. Search is broken. Social is worse.
You feel it every time you publish something meaningful — and it vanishes into the void.
SPP is the counter-move.
It lets content carry its own meaning, provenance, and context — wherever it goes.
We can either keep patching the broken system...
Or we can teach the web to speak for itself — semantically, contextually, and unambiguously.
The future of the web doesn’t need a new platform.
It needs a protocol.
- 📘 Canonical Specs –
/specs - 🧪 Working Examples –
/examples - 📤 Publishing Flows –
/flows - 📚 Visual Docs & Diagrams –
/docs
Start here: QUICKSTART.md
Browse all specs: spec-index.md
- JSON Schemas
- Consent & trust models
- Agent reconstruction logic
- End-to-end publishing flows
Full changelog: CHANGELOG.md
SPP follows Semantic Versioning 2.0.0 to ensure predictable updates and backward compatibility.
All SPP metadata includes a protocolVersion field that indicates which version of the protocol the content follows:
{
"protocolVersion": "1.0.0",
"id": "example-content",
"title": "Example Content"
}- MAJOR: Incremented for breaking changes that require content updates
- MINOR: Incremented for backward-compatible feature additions
- PATCH: Incremented for backward-compatible bug fixes
- Content published with protocol version
1.x.xwill remain compatible within the1.xseries - Agents and validators should gracefully handle minor version differences
- Major version changes may require content migration or updates
Use the built-in validation tool to check your drop metadata files against their JSON Schemas:
# Validate all site.config.json and pubs/**/meta.jsonld files
npm run validate
# Or run directly
node scripts/validate.js
# Validate specific files
node scripts/validate.js path/to/site.config.json path/to/meta.jsonldThe validator will:
- ✅ Print "All files pass validation" on success (exit code 0)
- ❌ Print "Validation failed" with detailed error information on failure (exit code 1)
- Show schema keyword, data path, and error message for each validation error
Works cross-platform (macOS/Linux/Windows) and is suitable for CI/CD pipelines.
Want to help fix the future of web publishing?
Start with CONTRIBUTING.md and GOVERNANCE.md
I taught the Web to speak.
Now it won’t shut up.