Technology

Solid is a W3C specification created by Tim Berners-Lee. At its core is a simple idea: give every person a Pod -- a typed, permission-controlled filesystem on the web. Every resource in a Pod has a URL and a content-type. The owner decides which applications and agents can read or write. Access is controlled at the protocol level, not by platform policy.

This is Enquire reconstructed for the modern web. Where Enquire was a local notebook, a Solid Pod is a web-addressable notebook -- yours to share selectively, yours to revoke access from, yours to migrate between providers without data loss. The applications that read your Pod are decoupled from the storage itself. You can switch clients without switching data.

Knowledge stored in your Pod only has value if it is accessible to the right tools. The Model Context Protocol is an open standard that defines how AI agents request and receive structured context from external sources. It is the interface between your knowledge base and any AI assistant you choose to use.

MCP is designed for local-first operation. A Blogmarks MCP server runs on your machine, queries a local vector index derived from your Pod, and returns semantically relevant context to the agent. The AI never accesses your Pod directly. Your data never leaves your device to reach Blogmarks servers. The query path is entirely local.

Schema.org is a collaborative vocabulary for structured data on the web. Founded in 2011 by the four largest search engines, it has been stable for over fifteen years, is used by more than 30% of web pages, and has never broken backward compatibility. It defines type hierarchies and properties that let any machine -- search engine, feed reader, AI assistant, knowledge graph -- understand content without custom parsing.

Structured data is how your content participates in the open web without depending on any single platform to interpret it. JSON-LD (JavaScript Object Notation for Linked Data) is the recommended serialization format: a script tag in your HTML that describes the content semantically. No special libraries. No build plugins. No runtime cost.

RDF (Resource Description Framework) is the W3C standard for describing resources as subject-predicate-object triples -- the smallest possible unit of meaning. Linked Data principles extend this: use URIs to name things, make those URIs dereferenceable, and link to other resources so that discovering one piece of data leads to more. WebID provides decentralized identity: a URI that identifies a person, resolvable to a profile document that describes them and their access permissions.

These are the foundations that make Solid Pods work. Data in a Pod is not just files -- it is typed, interlinked, and self-describing. A Blogmarks asset stored as RDF triples can be understood by any Linked Data client without prior knowledge of Blogmarks. WebID means your identity is not issued by a platform -- it is a URI you control.

Apache OpenDAL is a unified data access layer written in Rust. It provides a single Operator API for reading, writing, listing, and deleting across more than 40 storage backends -- including WebDAV (the protocol Solid Pods expose), local filesystem, S3-compatible stores, and in-memory backends for testing.

The value is backend portability. Application code uses one API. Switching from a local filesystem in development to a Solid Pod in production is a configuration change, not a code change. The abstraction does not leak Solid-specific concerns into application logic -- access control, WebID authentication, and RDF handling live in a thin layer above OpenDAL.