DJScorer

What it does

DJScorer Taste Guide turns a musical direction into artist exploration and a focused DJ practice session. Choose a genre, confirm a canonical Qloo music tag, and generate a plan. Qloo provides cultural relationships; the application turns them into practical exercises.

Why Qloo matters

The workflow grounds artist discovery in Qloo recommendations. Smooth-transition practice uses recommended artists, while discovery and energy goals can expand artists into related music tags. Provider affinities and the application's own practice instructions are labeled separately.

A tool for existing agents

A working MCP server exposes four schema-validated tools: taste_status, find_music_tags, build_practice_plan, and preview_practice_plan. Compatible agents can discover the tools, search for canonical tags, request a plan after tag confirmation, and use an explicit offline preview. The web interface and MCP server share the same planning routes. Practice instructions are deterministic; this project does not run an LLM or claim to evaluate a DJ's skill.

What was built during the hackathon

DJScorer existed as a largely unused, unpublished prototype before this event. The Qloo taste module was newly added October 13–14, 2026. Live adapter fixes, the standalone submission copy, and the MCP interface followed on October 15. This entry is a focused, independently runnable extraction. The earlier application was not all built during the hackathon.

Implementation and validation

The project uses TypeScript, JavaScript, React, Vite, Node.js, Express, the Qloo SDK, the MCP SDK, and Zod. The public demo runs on Netlify Functions with durable PostgreSQL-backed sessions and request accounting. Initial adapter validation used five bounded live Qloo requests and informed fixes to tag searches and artist subtype parsing. A separate hosted end-to-end check passed: House search, canonical tag confirmation, four real artist recommendations, and a 20-minute deterministic practice plan. That hosted check used two Qloo calls with no retries. Build, type checks, offline workflow/security tests, and MCP protocol tests also passed. The expanded artist-tag branch has mock-test coverage and was not part of the bounded live checks.

Demo boundaries

The offline example uses explicitly fictional artists, affinities, and venues, and never substitutes for failed live calls. No real event listings are bundled. Qloo affinities are aggregate relationships, not individual preference probabilities or attendance forecasts. The public source excludes secrets, private profiles, recordings, payment features, and the earlier app's private data.

Built With

• express.js
• javascript
• mcp
• netlify
• node.js
• postgresql
• qloo
• react
• typescript
• vite
• zod