← WebMCP Directory

Tool tree

paulkinlan.github.io

Developer ToolsTools verified
Website
Load real Wikipedia passage vectors into IndexedDB, then search them with the network off.
Explore capabilities

    page: Demo

    wikipedia_searchAction

    Semantic search over Wikipedia passage vectors stored in this browser; returns the five closest passages with cosine similarities.

    View tool JSON
    {
  "name": "wikipedia_search",
  "kind": "act",
  "impl": "declarative",
  "description": "Semantic search over Wikipedia passage vectors stored in this browser; returns the five closest passages with cosine similarities.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "query": {
        "type": "string",
        "description": "How do plants turn sunlight into food?"
      }
    },
    "required": [
      "query"
    ]
  },
  "page": "/idb-vector/demo/"
}

    page: Compute

    run_cosine_benchmarkAction

    Compare cosine scoring across IndexedDB, in-memory JavaScript, WASM-SIMD and WebGPU engines on a synthetic corpus; reports timings, storage costs and top-k agreement.

    View tool JSON
    {
  "name": "run_cosine_benchmark",
  "kind": "act",
  "impl": "declarative",
  "description": "Compare cosine scoring across IndexedDB, in-memory JavaScript, WASM-SIMD and WebGPU engines on a synthetic corpus; reports timings, storage costs and top-k agreement.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "count": {
        "type": "string",
        "enum": [
          "1000",
          "10000",
          "100000"
        ]
      },
      "dim": {
        "type": "string",
        "enum": [
          "32",
          "128",
          "384"
        ]
      }
    }
  },
  "page": "/idb-vector/demo/compute/"
}