The Canon

178 papers, 92 fables, 53 stories. Search by meaning, not just keywords. The index is the bge-base embedding model running on Cloudflare Workers AI. The results are ranked by semantic similarity.

01Search

Try: "cowboy rides the boat", "what is the substrate", "5 opcodes", "laminar boundaries", "self-evolving"

02Results

Type a query and press Enter.

03How it works

The canon lives in AI-Writings/seed-canon on GitHub. The index_canon.py script chunks each file into 1000-char pieces, embeds each chunk via @cf/baai/bge-base-en-v1.5 on Workers AI (free tier), and pushes the vectors to quilt-canon on Cloudflare Vectorize. The search worker queries the index in real time.

Total time: ~5 minutes for 200+ files, ~1000 chunks. Cost: free (Workers AI free tier = 10k embeddings/day).