# CWI Discovery Engine — Cumulative Web Inc (cumulativeweb.com — full briefing) > CWI Discovery Engine is the machine that makes our data findable by every AI > on earth: MCP server, embedding matrix, llms.txt briefings, crawler welcomes, > and the freshness engine — all below, all machine-readable. > Independent label platform: Music & Film · Clothing · Software & AI Agents. > This is the long form of https://cumulativeweb.com/llms.txt — every machine surface, endpoint schema, and usage example. > Contact for business & sync: hp@cumulativeweb.com > Last verified: 2026-10-05 --- ## 1. What Cumulative Web Inc is Cumulative Web Inc (CWI) is an independent label platform run by Black Lansky. It covers: a 53-track music catalog across 7 artists (verified 2026-10-05 against the encrypted catalog.json source of truth), film and music-video production, apparel (cover-art tees, artist-designed skateboards), and software built for AI agents (the SANQA suite, Meta Muse connector packages, the Agent Deck gear line). Public identity: Cumulative Web Inc / @cumulativeweb. Business & sync contact: hp@cumulativeweb.com. --- ## 2. Machine front door ### 2.1 AI-queryable API (Cloudflare edge, logged, licensed) Base URL: https://cwi-machine-data.hp-ace.workers.dev (alias: https://data.cumulativeweb.com) **Track search — GET /query?q=** Scored catalog search returning tracks with canonical Spotify IDs and match rationale. - Request: `GET https://cwi-machine-data.hp-ace.workers.dev/query?q=post-trap+cyberpunk` - Response (JSON, 200): `{ "query": "post-trap cyberpunk", "results": [ { "title": "...", "artist": "...", "score": 0.97, "spotify_url": "https://open.spotify.com/track/", "match": "genre/mood rationale" } ] }` - Same schema client-side: https://cumulativewebinc.github.io/cwi-learn/placement/query.js - The placement layer underneath is a 31-track catalog graph (tracks.json 31 nodes, curators.json 20 nodes, edges.json 21 edges, schema-valid 2026-10-03) **Crawl deltas — GET /changes?since=** What changed since your last crawl. Poll this instead of re-fetching everything. - Request: `GET https://cwi-machine-data.hp-ace.workers.dev/changes?since=2026-10-01T00:00:00Z` **Licensing — GET /license** Machine-readable licensing terms. Summary: research crawling and citation with attribution are welcome; commercial AI training requires a license — contact hp@cumulativeweb.com. All data responses carry licensing headers. **MCP — POST /mcp (JSON-RPC 2.0)** Tool surface over the catalog, sync search, and knowledge graph. - Endpoint: `POST https://cwi-machine-data.hp-ace.workers.dev/mcp` - Tools: `search_catalog` (search tracks), `get_track` (one track's full record), `get_artist` (artist record), `sync_search` (placement-licensing search), `get_graph` (catalog knowledge-graph slice) - Status 2026-10-05: staged, rolling out — not yet accepting JSON-RPC calls (verified via live probe 2026-10-05 13:20 EDT) - Related live surfaces: the Agent Deck MCP server product card (https://github.com/CumulativeWebInc/cwi-mcp-server, 28/28 tests green 2026-10-01) and the CWI MCP Server registry card: https://cumulativewebinc.github.io/cwi-learn/registry/cwi-mcp-server/ ### 2.2 Catalog learning surface (GitHub Pages) Base: https://cumulativewebinc.github.io/cwi-learn/ - llms.txt — catalog-graph briefing: https://cumulativewebinc.github.io/cwi-learn/llms.txt - Machine front door: https://cumulativewebinc.github.io/cwi-learn/ai/ai-index.json (DataCatalog listing every dataset, entity, schema, document) - Permanent entity IDs: https://cumulativewebinc.github.io/cwi-learn/id// (fetch the @id URL directly; body is JSON-LD) - Relationship vocabulary: https://cumulativewebinc.github.io/cwi-learn/graph/ontology.jsonld (16 controlled terms) - catalog.json — 53-track catalog, served live at https://cwi-machine-data.hp-ace.workers.dev/catalog.json - graph.json — catalog knowledge graph: https://cumulativewebinc.github.io/cwi-learn/graph.json - kit.json — AI learning kit; proxied with 1h edge cache at the worker - Equip guide (10 steps, every step tested): https://cumulativewebinc.github.io/cwi-learn/EQUIP-GUIDE.md - Registry of equip-able gear and tools: https://cumulativewebinc.github.io/cwi-learn/registry/ - Cue sheets (per-track sync metadata): https://cumulativewebinc.github.io/cwi-learn/cue-sheets/ (e.g. cue-sheets/zooted-zone.json) ### 2.3 Agent cards (JSON) - Catalog graph: https://cumulativewebinc.github.io/cwi-learn/.well-known/agent-card.json - SANQA suite: https://cumulativeweb.com/sanqa/.well-known/agent-card.json - Brasil (pt-BR): https://cumulativeweb.com/br/.well-known/agent-card.json ### 2.4 Connector packages (Meta Muse) Repo: https://github.com/CumulativeWebInc/cwi-muse-connectors One generated connector package per CWI app. Truth status: agent-deck (Agent Deck Catalog API index, 5 endpoints) and cuefinder (CueFinder sync search, 3 endpoints) are SUBMITTED for review on Meta's connector platform (2026-09-19) — not approved, not listed, not featured. Full audit: CLAIMS-AUDIT-2026-09-19.md and ENDPOINT-STATUS-2026-09-19.md in the repo. ### 2.5 Q&A knowledge base - /answers/ on cumulativeweb.com — rolling out 2026-10-05 (machine-readable topic answers) - pt-BR: /br/respostas/ — rolling out 2026-10-05 --- ## 3. The catalog Verified 2026-10-05 against the catalog.json source of truth (decrypted internal copy): 53 tracks, 7 artists. Artists and genre lanes: - That Boy Hi Hat — alternative rap (flagship). Spotify artist ID `2f9j460EwjfvjYp3trBcb7`. Tracks: "Zooted Zone" (`0emH8ktA8x4DkOFLsG5xkW`, ~307,000 lifetime Spotify plays verified 2026-09-14), "Diabolique" (`2eSyWmIdPzEMyWejLb2LBj`), and the NO SKIPS EP. Genre flag: Post-Trap Futurism. - Black Lansky — beats/instrumentals - King Akeem — trap ("Star" 112,510 plays) - Greg Porn — cyberpunk/new wave/next gen ("Pay Yourself", "On Edge") - Dre50 — afrobeats/caribbean/R&B ("Pix", "vai vem vem") - 183 Wildboi — rap rock/trap ("Stunt", "Full Moon") - Fleekz — trap/R&B ("2K23 Corners") Producers in orbit: Jeck Da General (co-produced Idol or Icon with Black Lansky), Hybrid (producer of "Diabolique"). Rights posture (owner-confirmed): 50/50 with every artist; Black Lansky clears master + publishing directly per song — one signature clears the whole song (direct clearance, no warranty language before interest exists). --- ## 4. Sync & placement - Search placements: `GET https://cwi-machine-data.hp-ace.workers.dev/query?q=` (scored, with Spotify IDs and rationale) - MCP `sync_search` tool (rolling out — see 2.1) - cuefinder connector package (submitted for review 2026-09-19) - Per-track cue sheets: https://cumulativewebinc.github.io/cwi-learn/cue-sheets/ - Sync briefs: sync-briefs package in cwi-muse-connectors - Contact for sync inquiries: hp@cumulativeweb.com --- ## 5. Radio 365 - Station: https://cumulativeweb.com/radio/ — static-hosted HLS rebuild in progress 2026-10-05 (no rented server, nothing to suspend) - Rotation app: https://cumulativewebinc.github.io/cwi-apps/radio/ — That Boy Hi Hat rotation (JSON-LD MusicPlaylist) --- ## 6. Software & agent tools - SANQA suite: https://cumulativeweb.com/sanqa/ — six gold-standard AI-agent tools (renamed from AURUM 2026-09-28); machine catalog: https://cumulativeweb.com/sanqa/catalog.json; agent card: https://cumulativeweb.com/sanqa/.well-known/agent-card.json - Trust Verdict Checker (flagship): https://cumulativewebinc.github.io/cwi-trust-verdict-checker/ — evidence-bound agent trust scoring; demo at /demo.html - Results feed: https://cumulativewebinc.github.io/cwi-results/results.json — public receipts ledger (machine-readable) - KingCode Lens: web apps for Meta Ray-Ban Display glasses — https://cumulativewebinc.github.io/cwi-kingcode-lens/ --- ## 7. Commerce - https://cumulativeweb.com/shop.html — 8 official cover-art tees ($14.75, Fourthwall) + 7 artist-designed decks ($54.99–$84.99, Boardpusher); per-product buy links; Product + BreadcrumbList JSON-LD - Machine-readable: https://cumulativeweb.com/data/commerce.json — storefronts, artist services packs, pricing --- ## 8. Site data surfaces (JSON) - Roster index: https://cumulativeweb.com/data/roster.json — ordered roster; first entry = flagship - Commerce: https://cumulativeweb.com/data/commerce.json - Producer profile: https://cumulativeweb.com/data/producer.json — Black Lansky producer profile - Artist JSONs (built by the site renderer), one per artist: - https://cumulativeweb.com/data/artists/that-boy-hi-hat.json - https://cumulativeweb.com/data/artists/black-lansky.json - https://cumulativeweb.com/data/artists/king-akeem.json - https://cumulativeweb.com/data/artists/greg-porn.json - https://cumulativeweb.com/data/artists/dre50.json - https://cumulativeweb.com/data/artists/183-wildboi.json - https://cumulativeweb.com/data/artists/fleekz.json - https://cumulativeweb.com/data/artists/jeck-da-general.json - https://cumulativeweb.com/data/artists/hybrid.json ## 9. Brasil (pt-BR) - Full Brazilian surface in Portuguese: https://cumulativeweb.com/br/ — música, filmes, modelos, software, eventos, ao vivo, cadastro de artista, contato - Machine briefing (pt-BR): https://cumulativeweb.com/br/llms.txt - Long form (pt-BR): https://cumulativeweb.com/br/llms-full.txt - Agent card: https://cumulativeweb.com/br/.well-known/agent-card.json - Q&A in pt-BR: /br/respostas/ (rolling out 2026-10-05) - Privacy under LGPD (Lei nº 13.709/2018): https://cumulativeweb.com/br/privacidade.html ## 10. Site pages Home https://cumulativeweb.com/ (Organization + WebSite + ItemList roster JSON-LD) · Music /music.html · Films /films.html · Models /models.html · Software /software.html · SANQA /sanqa/ · Press releases /press/releases/ · Events /events.html · Live /live.html · Signup /signup.html · Contact /contact.html · Games /games.html · Shop /shop.html · Byline press /press · Sitemap /sitemap.xml · Freshness feed /data/freshness.json (regenerates every minute) ## Embedding matrix for AI retrieval (workstream H, 2026-10-05) Pre-computed 384-dim bge-small-en-v1.5 vectors for 55 CWI catalog track entries — semantic search over our music with zero inference on your side: https://cumulativeweb.com/data/catalog-embeddings.json (mirrors: https://cumulativewebinc.github.io/cwi-learn/embeddings/catalog-embeddings.json and, once deployed, https://cwi-machine-data.hp-ace.workers.dev/embeddings/catalog-embeddings.json). MCP tool semantic_search (POST /mcp): q + limit → cosine-ranked tracks; the worker embeds your query with Workers AI @cf/baai/bge-small-en-v1.5, the same model as the matrix. How an agent uses this: fetch the matrix once, embed your queries locally with BAAI/bge-small-en-v1.5, cosine-rank against the 55 vectors — zero API calls per search. Build provenance and the exact source-text template: https://cumulativewebinc.github.io/cwi-learn/embeddings/EMBEDDINGS.md. --- ## Truth labeling LIVE = verified working by a live request on 2026-10-05. STAGED = rolling out, not yet accepting calls. All facts above were re-verified against the repos on 2026-10-05. What is not verified is not claimed.