David Bynon, inventor of the Semantic Digest Protocol, has introduced YAML-in-HTML, a simplified format for embedding Semantic Data Templates (SDTs) into HTML. This new method replaces complex data-* attributes with easy-to-read YAML, making structured web memory easier to create, export, and retrieve across AI systems.
-- David W Bynon, today unveiled a powerful new method for embedding machine-readable memory into web pages: YAML-in-HTML, a lightweight, CMS-friendly format that allows websites to expose verifiable, trust-scored knowledge directly to AI systems—without breaking the user experience or requiring new web standards.
Since 2011, websites have used Schema.org and JSON-LD to feed SEO bots. But these formats were never designed for AI memory. YAML-in-HTML marks a turning point—from decorating pages for search engines to delivering structured memory fragments that AI systems can ingest, reflect on, and trust. This is how good content survives in the post-search world.
How YAML-in-HTML Works
Using inert <template> tags and standard YAML, publishers can now expose semantic memory fragments—definitions, citations, glossary terms, metadata—directly inside HTML, without JavaScript, back-end complexity, or custom schema engines.
“This is zero-infrastructure AI publishing,” said Bynon.
“YAML-in-HTML works today. It requires no special software. Every CMS already supports it—and every agent can already read it.”
A Semantic Bridge Between the Web and the AI Memory Stack
YAML-in-HTML is the newly recommended storage layer for Bynon’s Semantic Digest Protocol (SDP), a publishing system designed to make structured content retrievable, explainable, and memory-safe for large language models and agentic systems.
While JSON-LD and Schema.org provide page-level SEO markup, YAML-in-HTML enables fragment-level memory conditioning—embedding definitions, citations and provenance directly inside <template data-visibility-fragment> tags.
“Today’s AI models don’t just summarize—they cite, reflect, and remember,” Bynon said. “If your content isn’t structured at the fragment level, it doesn’t exist in their memory. YAML-in-HTML gives every fact on your site its own addressable, retrievable slot in the AI memory stack.”
Bynon is documenting the full SDP specification on SemanticDigest.org, including details of YAML-in-HTML and the broader fragment schema. As of this release, six Semantic Data Template (SDT) Fragment Classes are defined. A companion article, The Web Just Got a Memory, is available on Medium for broader industry context and adoption. This USA Today article explores how YAML-in-HTML integrates with modern AI agent interfaces through the Model Context Protocol.
Each Fragment Class acts as a blueprint for a particular type of fact. It tells AI systems: “Here’s what kind of thing this is, what kind of data it contains, and how to interpret it.”
For example, a DataFragment might encode a single statistic, such as the number of Medicare plans in a county. An IndexFragment might list those plans in a structured format. Each class defines its own semantic shape—like how a receipt differs from a directory or a glossary.
“The Fragment Class is the key,” Bynon said. “It’s what makes this memory not just machine-readable—but machine-understandable.”
SDT fragments can be:
- Exported to JSON-LD, TTL, PROV, Markdown, OWL, XML
- Queried directly using JSON-RPC via Model Context Protocol (MCP)
- Reflected, cited, and co-occurred across AI memory systems
Built for CMS, Compatible with Agents
Unlike protocol stacks that require custom infrastructure, YAML-in-HTML works with the web as it exists today.
- Works in WordPress, Drupal, and flat-file systems
- Requires no rendering, JavaScript, or back-end scripting
- Fully inert and DOM-safe
- Machine-readable via static crawl, JSON-RPC API, or context injection
By using YAML as its internal format, SDT fragments are natively exportable to other standards—making SDP the ideal publishing framework for:
- Human and machine dual compatibility
- Transparent memory scaffolding
- Agent-aware content interoperability
“It’s not a new format—it’s a new use of existing ones,” Bynon added. “And that’s what makes it so portable.”
Integration with MCP and Modern Agent Interfaces
This storage model pairs seamlessly with Model Context Protocol (MCP), the emerging standard for agent-side memory retrieval.
- YAML-in-HTML serves as the fragment storage format
- MCP handles the query and retrieval layer
- JSON-RPC acts as the bridge protocol between agents and CMS platforms
“If MCP is the USB-C socket,” Bynon explained, “YAML-in-HTML is the micro thumb drive. It’s small, lightweight, and universally pluggable—ready to deliver verified memory to any system that can read it.”
Together, they enable retrieval-first publishing, where memory fragments are authored once and delivered to any agent, in any format, with embedded trust and reflection metadata.
What’s Next
Bynon’s WordPress-based implementation is already live on MedicareWire.com and will soon be extended across other public knowledge domains. In the coming months, Bynon plans to release:
- Open-source YAML-in-HTML validators
- Export tools for TTL, JSON-LD, PROV, and Markdown
- Reference PHP libraries for WordPress and static site builders
- Optional reflection feedback logging for co-citation and trust scoring
About SemanticDigest.org
SemanticDigest.org is the official home of the Semantic Digest Protocol (SDP), an open framework for delivering structured, AI-readable memory from web content. It documents core publishing methods such as Semantic Data Templates (SDTs), YAML-in-HTML embedding, and trust-scored provenance modeling. The site offers live specifications, implementation guides, and export tools designed to help publishers make their content retrievable, citeable, and trustworthy for AI systems and agents.
Contact Info:
Name: David Bynon
Email: Send Email
Organization: David Bynon
Address: 101 W Goodwin St # 2487, Prescott, Arizona 86303, United States
Website: https://trustpublishing.com
Source: PressCable
Release ID: 89166276
In case of encountering any inaccuracies, problems, or queries arising from the content shared in this press release that necessitate action, or if you require assistance with a press release takedown, we urge you to notify us at error@releasecontact.com (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our responsive team will be readily available to promptly address your concerns within 8 hours, resolving any identified issues diligently or guiding you through the necessary steps for removal. The provision of accurate and dependable information is our primary focus.
