{
  "id": 11248333,
  "title": "I gave my coding agent a sense of taste. It picks restaurants from my music.",
  "url": "https://urgent.news/2026/10/01/i-gave-my-coding-agent-a-sense-of-taste-it-picks-restaurants-from-my",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-01T17:57:06.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/haoli/i-gave-my-coding-agent-a-sense-of-taste-it-picks-restaurants-from-my-music-4o1"
  },
  "original_language": "en",
  "account": "An innovative coding agent was given a sense of taste, enabling it to make restaurant recommendations based on a user's musical preferences. The Qloo Taste API, containing 250 million cultural entities, acts as the foundation for this unique capability. The API links various cultural elements such as music, movies, restaurants, brands, and destinations through a graph, allowing the agent to draw connections between disparate domains.\n\nTo make this possible, Qloo developed qloo-taste-mcp, a server that provides four tools for the agent to utilize. These tools include qloo_search for searching the taste graph using entity IDs, qloo_recommend for generating recommendations based on those IDs, qloo_trending for discovering current trends in specific categories, and qloo_tags for searching a taxonomy of genres, cuisines, and travel themes.\n\nThe most intriguing aspect is the cross-domain transfer of taste: by connecting musical preferences with culinary recommendations. For instance, if a user loves Taylor Swift and Wes Anderson films, the system can find a suitable restaurant in LA by searching for seeds and then using qloo_recommend with the appropriate entity ID.\n\nNotably, the system requires no third-party dependencies, relying solely on Python's standard library. To use the tools, users must export their QLOO_API_KEY and run the qloo_taste_mcp.server script. A live demo called Taste Match is available, allowing users to input their liked items, select a category, and receive personalized recommendations.\n\nThe code is MIT licensed and built for the Qloo Agentic Hackathon, with the source code available on GitHub. With this technology, an agent could suddenly begin developing opinions about ramen, reflecting the influence of its newfound taste sense.",
  "summary": "My coding agent can refactor a thousand lines without breaking a sweat, but ask it \"where should I take a date Friday night\" and it's useless. It knows my code. It knows nothing about my taste. Qloo's Taste API sits on 250M+ cultural entities — music, movies, restaurants, brands, destinations — with a graph of how tastes connect. So I built qloo-taste-mcp : an MCP server that hands all of that to…",
  "key_points": [
    "Qloo coding agent gains taste sense",
    "Qloo Taste API connects 250M cultural entities",
    "Agent recommends restaurants based on music preferences"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}