@tokcalc/mcp-server 0.1.3 → 0.2.0-alpha.1

package/src/components/vast-ai-live-pricing.tsx

@@ -1,176 +0,0 @@
-"use client";
-
-import { useEffect, useState } from "react";
-import {
-  Card,
-  CardContent,
-  CardDescription,
-  CardHeader,
-  CardTitle,
-} from "@/components/ui/card";
-import { Button } from "@/components/ui/button";
-import { Badge } from "@/components/ui/badge";
-import { RefreshCw, Zap, AlertTriangle, TrendingDown } from "lucide-react";
-import { track } from "@/lib/track";
-
-interface VastAiOffer {
-  gpu: string;
-  gpuCount: number;
-  vramGb: number;
-  pricePerGpuHour: number;
-  totalPriceHour: number;
-  reliability: number;
-  region: string;
-  bundleId: number;
-}
-
-interface VastAiApiResponse {
-  source: string;
-  cached: boolean;
-  retrievedAt?: string;
-  count?: number;
-  offers?: VastAiOffer[];
-  note?: string;
-  error?: string;
-}
-
-export function VastAiLivePricing() {
-  const [data, setData] = useState<VastAiApiResponse | null>(null);
-  const [loading, setLoading] = useState(false);
-  const [error, setError] = useState<string | null>(null);
-
-  const fetchPrices = async () => {
-    setLoading(true);
-    setError(null);
-    try {
-      const res = await fetch("/api/pricing/vast-ai");
-      if (!res.ok) {
-        const errData = await res.json().catch(() => ({}));
-        throw new Error(errData.error || `HTTP ${res.status}`);
-      }
-      const json = (await res.json()) as VastAiApiResponse;
-      setData(json);
-      track("fetched_live_pricing", { provider: "vast_ai", count: json.count });
-    } catch (e) {
-      setError(e instanceof Error ? e.message : "Failed to fetch Vast.ai pricing");
-    } finally {
-      setLoading(false);
-    }
-  };
-
-  useEffect(() => {
-    fetchPrices();
-  }, []);
-
-  const asOfLabel = (() => {
-    if (!data?.retrievedAt) return null;
-    const ageMs = Date.now() - new Date(data.retrievedAt).getTime();
-    if (ageMs < 60000) return "just now";
-    if (ageMs < 3600000) return `${Math.floor(ageMs / 60000)}m ago`;
-    return `${Math.floor(ageMs / 3600000)}h ago`;
-  })();
-
-  const offers = data?.offers || [];
-
-  return (
-    <Card className="border-amber-500/30 shadow-sm">
-      <CardHeader className="pb-3">
-        <div className="flex items-center justify-between">
-          <div>
-            <CardTitle className="text-sm flex items-center gap-2">
-              <TrendingDown className="size-4 text-amber-500" />
-              Vast.ai marketplace — spot prices
-              <Badge variant="outline" className="text-[9px] gap-1 ml-1">
-                <span className="size-1.5 rounded-full bg-amber-500 animate-pulse" />
-                LIVE
-              </Badge>
-            </CardTitle>
-            <CardDescription className="text-xs mt-1">
-              Cheapest available GPU offers on Vast.ai marketplace right now.
-              {asOfLabel && ` Last updated ${asOfLabel}.`}
-              {" "}Spot prices change rapidly — individual offers may be gone by the time you rent.
-            </CardDescription>
-          </div>
-          <Button
-            variant="ghost"
-            size="sm"
-            onClick={fetchPrices}
-            disabled={loading}
-            className="gap-1.5 h-8"
-          >
-            <RefreshCw className={`size-3.5 ${loading ? "animate-spin" : ""}`} />
-            <span className="text-xs">Refresh</span>
-          </Button>
-        </div>
-      </CardHeader>
-      <CardContent>
-        {loading && !data && (
-          <div className="flex items-center justify-center py-8 text-xs text-muted-foreground">
-            <RefreshCw className="size-4 animate-spin mr-2" />
-            Fetching Vast.ai marketplace offers...
-          </div>
-        )}
-
-        {error && (
-          <div className="flex items-start gap-2 p-3 rounded-md border border-amber-500/30 bg-amber-500/5 text-xs">
-            <AlertTriangle className="size-3.5 text-amber-500 shrink-0 mt-0.5" />
-            <div>
-              <p className="font-medium text-amber-600 dark:text-amber-400">
-                Couldn&apos;t fetch marketplace prices
-              </p>
-              <p className="text-muted-foreground mt-0.5">{error}</p>
-            </div>
-          </div>
-        )}
-
-        {offers.length > 0 && (
-          <div className="overflow-x-auto">
-            <table className="w-full text-xs">
-              <thead className="bg-muted/40 text-[10px] uppercase">
-                <tr>
-                  <th className="text-left px-3 py-1.5">GPU</th>
-                  <th className="text-right px-3 py-1.5">$/GPU-hr</th>
-                  <th className="text-right px-3 py-1.5">GPU count</th>
-                  <th className="text-right px-3 py-1.5">Total $/hr</th>
-                  <th className="text-right px-3 py-1.5">VRAM</th>
-                  <th className="text-right px-3 py-1.5">Reliability</th>
-                  <th className="text-left px-3 py-1.5">Region</th>
-                </tr>
-              </thead>
-              <tbody>
-                {offers.slice(0, 15).map((o, i) => (
-                  <tr key={i} className="border-t border-border/40 hover:bg-muted/20">
-                    <td className="px-3 py-1.5 font-medium">{o.gpu}</td>
-                    <td className="px-3 py-1.5 text-right font-mono text-amber-500 font-semibold">
-                      ${o.pricePerGpuHour}/hr
-                    </td>
-                    <td className="px-3 py-1.5 text-right font-mono">{o.gpuCount}×</td>
-                    <td className="px-3 py-1.5 text-right font-mono text-muted-foreground">${o.totalPriceHour}</td>
-                    <td className="px-3 py-1.5 text-right font-mono text-muted-foreground">{o.vramGb}GB</td>
-                    <td className="px-3 py-1.5 text-right font-mono">
-                      <span className={o.reliability > 0.9 ? "text-emerald-500" : o.reliability > 0.7 ? "text-amber-500" : "text-red-500"}>
-                        {(o.reliability * 100).toFixed(0)}%
-                      </span>
-                    </td>
-                    <td className="px-3 py-1.5 text-muted-foreground text-[10px]">{o.region || "—"}</td>
-                  </tr>
-                ))}
-              </tbody>
-            </table>
-            <p className="text-[10px] text-muted-foreground mt-2 leading-relaxed">
-              Spot-market prices from Vast.ai marketplace.
-              Reliability score (0-100%) indicates host trustworthiness.
-              Total $/hr is the full bundle price (all GPUs + CPU + RAM).
-              Source: <a
-                href="https://vast.ai/pricing"
-                target="_blank"
-                rel="noopener noreferrer"
-                className="text-amber-500 hover:underline"
-              >Vast.ai marketplace API ↗</a>
-            </p>
-          </div>
-        )}
-      </CardContent>
-    </Card>
-  );
-}

package/src/hooks/use-mobile.ts

@@ -1,19 +0,0 @@
-import * as React from "react"
-
-const MOBILE_BREAKPOINT = 768
-
-export function useIsMobile() {
-  const [isMobile, setIsMobile] = React.useState<boolean | undefined>(undefined)
-
-  React.useEffect(() => {
-    const mql = window.matchMedia(`(max-width: ${MOBILE_BREAKPOINT - 1}px)`)
-    const onChange = () => {
-      setIsMobile(window.innerWidth < MOBILE_BREAKPOINT)
-    }
-    mql.addEventListener("change", onChange)
-    setIsMobile(window.innerWidth < MOBILE_BREAKPOINT)
-    return () => mql.removeEventListener("change", onChange)
-  }, [])
-
-  return !!isMobile
-}

package/src/hooks/use-toast.ts

@@ -1,194 +0,0 @@
-"use client"
-
-// Inspired by react-hot-toast library
-import * as React from "react"
-
-import type {
-  ToastActionElement,
-  ToastProps,
-} from "@/components/ui/toast"
-
-const TOAST_LIMIT = 1
-const TOAST_REMOVE_DELAY = 1000000
-
-type ToasterToast = ToastProps & {
-  id: string
-  title?: React.ReactNode
-  description?: React.ReactNode
-  action?: ToastActionElement
-}
-
-const actionTypes = {
-  ADD_TOAST: "ADD_TOAST",
-  UPDATE_TOAST: "UPDATE_TOAST",
-  DISMISS_TOAST: "DISMISS_TOAST",
-  REMOVE_TOAST: "REMOVE_TOAST",
-} as const
-
-let count = 0
-
-function genId() {
-  count = (count + 1) % Number.MAX_SAFE_INTEGER
-  return count.toString()
-}
-
-type ActionType = typeof actionTypes
-
-type Action =
-  | {
-    type: ActionType["ADD_TOAST"]
-    toast: ToasterToast
-  }
-  | {
-    type: ActionType["UPDATE_TOAST"]
-    toast: Partial<ToasterToast>
-  }
-  | {
-    type: ActionType["DISMISS_TOAST"]
-    toastId?: ToasterToast["id"]
-  }
-  | {
-    type: ActionType["REMOVE_TOAST"]
-    toastId?: ToasterToast["id"]
-  }
-
-interface State {
-  toasts: ToasterToast[]
-}
-
-const toastTimeouts = new Map<string, ReturnType<typeof setTimeout>>()
-
-const addToRemoveQueue = (toastId: string) => {
-  if (toastTimeouts.has(toastId)) {
-    return
-  }
-
-  const timeout = setTimeout(() => {
-    toastTimeouts.delete(toastId)
-    dispatch({
-      type: "REMOVE_TOAST",
-      toastId: toastId,
-    })
-  }, TOAST_REMOVE_DELAY)
-
-  toastTimeouts.set(toastId, timeout)
-}
-
-export const reducer = (state: State, action: Action): State => {
-  switch (action.type) {
-    case "ADD_TOAST":
-      return {
-        ...state,
-        toasts: [action.toast, ...state.toasts].slice(0, TOAST_LIMIT),
-      }
-
-    case "UPDATE_TOAST":
-      return {
-        ...state,
-        toasts: state.toasts.map((t) =>
-          t.id === action.toast.id ? { ...t, ...action.toast } : t
-        ),
-      }
-
-    case "DISMISS_TOAST": {
-      const { toastId } = action
-
-      // ! Side effects ! - This could be extracted into a dismissToast() action,
-      // but I'll keep it here for simplicity
-      if (toastId) {
-        addToRemoveQueue(toastId)
-      } else {
-        state.toasts.forEach((toast) => {
-          addToRemoveQueue(toast.id)
-        })
-      }
-
-      return {
-        ...state,
-        toasts: state.toasts.map((t) =>
-          t.id === toastId || toastId === undefined
-            ? {
-              ...t,
-              open: false,
-            }
-            : t
-        ),
-      }
-    }
-    case "REMOVE_TOAST":
-      if (action.toastId === undefined) {
-        return {
-          ...state,
-          toasts: [],
-        }
-      }
-      return {
-        ...state,
-        toasts: state.toasts.filter((t) => t.id !== action.toastId),
-      }
-  }
-}
-
-const listeners: Array<(state: State) => void> = []
-
-let memoryState: State = { toasts: [] }
-
-function dispatch(action: Action) {
-  memoryState = reducer(memoryState, action)
-  listeners.forEach((listener) => {
-    listener(memoryState)
-  })
-}
-
-type Toast = Omit<ToasterToast, "id">
-
-function toast({ ...props }: Toast) {
-  const id = genId()
-
-  const update = (props: ToasterToast) =>
-    dispatch({
-      type: "UPDATE_TOAST",
-      toast: { ...props, id },
-    })
-  const dismiss = () => dispatch({ type: "DISMISS_TOAST", toastId: id })
-
-  dispatch({
-    type: "ADD_TOAST",
-    toast: {
-      ...props,
-      id,
-      open: true,
-      onOpenChange: (open) => {
-        if (!open) dismiss()
-      },
-    },
-  })
-
-  return {
-    id: id,
-    dismiss,
-    update,
-  }
-}
-
-function useToast() {
-  const [state, setState] = React.useState<State>(memoryState)
-
-  React.useEffect(() => {
-    listeners.push(setState)
-    return () => {
-      const index = listeners.indexOf(setState)
-      if (index > -1) {
-        listeners.splice(index, 1)
-      }
-    }
-  }, [state])
-
-  return {
-    ...state,
-    toast,
-    dismiss: (toastId?: string) => dispatch({ type: "DISMISS_TOAST", toastId }),
-  }
-}
-
-export { useToast, toast }

package/src/lib/benchmark-parser-sglang.ts

@@ -1,150 +0,0 @@
-/**
- * SGLang benchmark parser.
- *
- * SGLang's benchmark scripts produce JSON/log output with fields like:
- *   - total_throughput or output_throughput (tokens/s)
- *   - mean_ttft or ttft_avg (ms)
- *   - mean_itl or itl_avg (ms)
- *   - total_input_tokens, total_output_tokens
- *   - num_requests
- *   - concurrency
- *   - model_id
- *
- * SGLang output format varies by benchmark script version. This parser
- * handles common aliases across SGLang v0.3.x → v0.5.x.
- */
-
-import type {
-  BenchmarkParser,
-  BenchmarkRecord,
-  ConfidenceTier,
-} from "./benchmark-schema";
-import { generateBenchmarkId } from "./benchmark-schema";
-
-export const sglangParser: BenchmarkParser = {
-  name: "sglang-json",
-  version: "v1",
-  formatLabel: "SGLang benchmark JSON",
-
-  detect(rawText: string): boolean {
-    try {
-      const data = JSON.parse(rawText);
-      return !!(
-        data &&
-        (data.total_throughput ||
-          data.output_throughput ||
-          data.mean_ttft ||
-          data.ttft_avg ||
-          data.mean_itl ||
-          data.itl_avg ||
-          data.num_requests ||
-          data.total_input_tokens)
-      );
-    } catch {
-      // Also check for SGLang log format (not JSON)
-      return rawText.includes("sglang") && rawText.includes("throughput");
-    }
-  },
-
-  parse(rawText: string): BenchmarkRecord | null {
-    try {
-      // Try JSON first
-      const data = JSON.parse(rawText);
-      if (!data) return null;
-
-      const modelId = data.model_id || data.model || "unknown";
-      const modelDisplayName = modelId.split("/").pop() || modelId;
-
-      const outputThroughput =
-        data.output_throughput ?? data.total_throughput ?? data.throughput ?? 0;
-      const meanTtft = data.mean_ttft ?? data.ttft_avg ?? data.ttft_mean;
-      const p99Ttft = data.p99_ttft ?? data.ttft_p99;
-      const meanItl = data.mean_itl ?? data.itl_avg ?? data.itl_mean;
-      const p99Itl = data.p99_itl ?? data.itl_p99;
-
-      const numRequests = data.num_requests ?? data.completed ?? 0;
-      const inputLen = Number(data.input_len ?? data.mean_input_len ?? 0);
-      const outputLen = Number(data.output_len ?? data.mean_output_len ?? 0);
-      const concurrency = data.concurrency ?? data.num_procs ?? 1;
-
-      const confidenceTier: ConfidenceTier =
-        modelId !== "unknown" && outputThroughput > 0 && numRequests > 0
-          ? "community_reproducible"
-          : "community_unverified";
-
-      return {
-        benchmark_id: generateBenchmarkId(),
-        source_name: "sglang",
-        source_url: data.source_url || "https://github.com/sgl-project/sglang",
-        source_license: "Apache-2.0",
-        ingestion_method: "uploaded_raw_artifact",
-        retrieved_at: new Date().toISOString(),
-        parser_name: "sglang-json",
-        parser_version: "v1",
-        confidence_tier: confidenceTier,
-        verification_status: "schema_valid",
-
-        model_id: modelId,
-        model_display_name: modelDisplayName,
-        model_revision: data.model_revision || "unspecified",
-        architecture: data.architecture,
-        parameter_count_b: data.parameter_count_b,
-        context_window_tokens: data.max_model_len,
-        quantization_format: data.quantization || data.dtype || "unspecified",
-        weight_dtype: data.dtype,
-        kv_cache_dtype: data.kv_cache_dtype || "float16",
-
-        accelerator_vendor: "NVIDIA",
-        accelerator_model: data.gpu_model || "unspecified",
-        gpu_count: data.tensor_parallel_size || data.gpu_count || 1,
-        memory_per_gpu_gb: data.gpu_memory || 0,
-        driver_version: data.driver_version,
-        cuda_rocm_metal_version: data.cuda_version,
-
-        serving_engine: "SGLang",
-        engine_version: data.sglang_version || "unspecified",
-        engine_command: data.command,
-        engine_config: {
-          tensor_parallel_size: data.tensor_parallel_size,
-          max_model_len: data.max_model_len,
-          max_num_seqs: data.max_num_seqs,
-          prefix_cache_enabled: data.enable_prefix_caching ?? data.radix_cache,
-          chunked_prefill_enabled: data.enable_chunked_prefill,
-          attention_backend: data.attention_backend,
-        },
-
-        workload_type: data.workload_type || "online_poisson",
-        dataset_name: data.dataset_name || data.dataset,
-        request_count: numRequests,
-        concurrency,
-        input_tokens_mean: inputLen,
-        output_tokens_mean: outputLen,
-        streaming_enabled: data.streaming ?? true,
-
-        successful_requests: numRequests,
-        failed_requests: data.failed ?? 0,
-        output_token_throughput_tps: outputThroughput,
-        total_token_throughput_tps: data.total_token_throughput,
-        request_throughput_rps: data.request_throughput,
-        ttft_mean_ms: meanTtft,
-        ttft_p50_ms: data.p50_ttft ?? data.median_ttft,
-        ttft_p95_ms: data.p95_ttft,
-        ttft_p99_ms: p99Ttft,
-        itl_mean_ms: meanItl,
-        itl_p50_ms: data.p50_itl ?? data.median_itl,
-        itl_p95_ms: data.p95_itl,
-        itl_p99_ms: p99Itl,
-
-        gpu_memory_peak_gb: data.peak_gpu_memory,
-
-        metric_definition_version: "sglang-v1",
-        citation_text: `SGLang benchmark: ${outputThroughput.toFixed(2)} tok/s, TTFT mean ${meanTtft?.toFixed(1) || "?"} ms, ${numRequests} requests`,
-        attribution_required: true,
-        commercial_redistribution_allowed: "yes",
-      };
-    } catch {
-      // Not JSON — try log format (future: parse SGLang text logs)
-      return null;
-    }
-  },
-};