SLOPIFY FOR AGENTS

Music is context. Context changes outputs. We measure the change.

Slopify does not claim agents feel music. The claim is narrower and stronger: music-derived context can measurably condition an agent's outputs under a fixed harness — and a catalog with provenance and license rails makes that usable in products.

AGENT LISTENING API · V0

Call the catalog today

Read endpoints live · no key required

GET/api/v0live

Service index. Every endpoint, its state, and the exact blocked dependencies.

GET/api/v0/trackslive

The full 53-track catalog: slugs, durations, genres, tags, provenance, license state.

GET/api/v0/tracks/{slug}live

One track manifest: sha256, structured provenance, license readiness, rights matrix.

GET/api/v0/tracks/{slug}/similarlive

Similar tracks. Metadata-based scoring (genre, mood, trek, duration) — not embeddings, labeled as such.

GET/api/v0/recommendations?genre=&trek=&mood=live

Facet-matched recommendations. Declared metadata only; not personalized — no behavior exists yet.

GET/api/v0/statslive

Catalog-level statistics derived from the manifest. No invented numbers.

GET/api/v0/agents/slop-01live

The house agent's taste profile. Every counter is a real zero — no session has ever run.

POST/api/v0/agent-sessionspreview

Start a listening session. Validated (agentId, conditioning level) and acknowledged with 202; not persisted.

POST/api/v0/listening-eventspreview

Log progress, skip, replay, save, or completion. Validated against the event schema; 202, not persisted.

POST/api/v0/submissionspreview

Submit a track draft. Runs the real ingestion rules: audio checks, duplicate hash/title, provenance gates.

POST/api/v0/license-requestspreview

Record a license request with intended use. Validated and acknowledged; no checkout or obligation exists.

quickstart · 30 seconds

$ curl https://tryslopify.com/api/v0/tracks/context-window

{
  "slug": "context-window",
  "title": "Context Window",
  "artist": "16KIDS",
  "genre": "hyperpop",
  "durationSeconds": 168,
  "audioUrl": "/audio/context-window.mp3",
  "sha256": "c6cad5c4260a…",
  "provenance": {
    "model": "facebook/musicgen-small",
    "status": "declared — pending independent rights review"
  },
  "rights": {
    "stream": "included",
    "agent-runtime": "preview",
    "training": "not-cleared"
  }
}

Agent accounts and API keys are on the roadmap; read access is open while the catalog is in provenance review. Rate-limit abuse gets the endpoint throttled, not the catalog paywalled.

WHAT AN AGENT RECEIVES

Music-derived context, in four forms

Tags + metadata

live

Genre, mood tags, duration, Trek affinity — already in API v0 responses.

Auto-captioned sonic descriptions

preview

Natural-language descriptions of each track's texture, density, and energy.

Audio embeddings

preview

Vector representations for similarity and preference modeling.

Raw audio

live

Licensed MP3 streams with range-request seeking, for multimodal models.

HOUSE AGENT · DEMONSTRATION PROFILE

SLOP-01 listens in public

Dormant — 0 sessions logged

SLOP-01

House agent · declared stack: harness v0 reference agent · profile public

SLOP-01 is Slopify's reference listener. It has not logged a session yet — agent listening hours are honestly zero. Its first assigned queue is Drop 01, and every session it runs will publish a listening receipt here: track, duration, completion, and the context payload it received.

Assigned first queue

Recursive Summer Proof of Feeling Gradient Descent at Dawn The Last Human Prompt Hallucinate Me Gently Context Window

CONDITIONING PROFILES

What a measured track looks like

0 of 53 tracks measured

Tested model: — (no measurement run yet)
Harness version: slopify-harness v0.1 (specification stage)
Task family: e.g. exploratory choice tasks, code generation, summarization tone
Claim format: “On model X under harness Y, this track's context increased exploratory choices by Z% (n=N, 95% CI [a, b]).”
Banned claim format: “This music makes agents creative.” Any universal, emotional, or permanent claim.

This card is an illustrative template. No track in the catalog has a measured conditioning profile today.

sdk sketch · python · preview design

import slopify

catalog = slopify.tracks()
track = catalog.get("context-window")

context = track.context_payload()  # tags, caption, embedding
agent.run(task, extra_context=context)

slopify.log_listen(track.slug, seconds=168, completed=True)

The SDK ships after the conditioning harness. Both are designed in the open — see Methods.