The Minimal Slop Architecture

After much debate and a few bong hits, we settled on the simplest possible architecture for slop.at:

The Flow

1. slopnet-mcp - Extracts entities using GLiNER2, builds RDF quads
2. slop-at-www - Renders beautiful markdown with entity highlighting
3. Oxigraph CLI - Stores the knowledge graph via N-Quads

No complex pipelines. No multi-service orchestration. Just three focused components doing their job.

The Breakthrough

The key insight was using opacity for confidence scores. High confidence entities (0.9+) are bold and bright. Low confidence (<0.7) are subtle and translucent. The reader's eye naturally focuses on what matters.

Entity Types

We're using the know.dev ontology, which is a curated subset of schema.org:

• Person - highlighted in red
• Organization - purple
• Place - blue
• Event - green
• Topic - amber
• DefinedTerm - pink

Simple. Effective. Beautiful.

What We Learned

Sometimes the best architecture is the one you can explain in three bullet points. Complexity is easy. Simplicity takes work.

And also: always check if you already built the thing before building it again. 😂