mcp-tollbooth 0.1.0 → 0.2.0

package/dist/auditor.js

@@ -1,5 +1,5 @@
 import { categorize, estimateTokens } from "./heuristics.js";
-import { classifyServer, fetchGithubRepoInfo, fetchNpmPackageInfo } from "./registryInfo.js";
+import { classifyServer, fetchGithubRepoInfo, fetchNpmPackageInfo, KNOWN_UVX_SERVER_REPOS } from "./registryInfo.js";
 const OFFICIAL_SCOPES = ["@modelcontextprotocol", "@anthropic-ai", "@upstash", "@notionhq", "@playwright"];
 function computeTrustScore(params) {
     const notes = [];
@@ -72,10 +72,17 @@ export async function auditConfigFiles(configFiles) {
         const { kind, packageName, rawCommandLine } = classifyServer(rawCfg);
         let packageInfo;
         let githubLookup;
-        if (packageName) {
+        if (packageName && kind === "npx-package") {
             packageInfo = (await fetchNpmPackageInfo(packageName)) ?? { name: packageName };
             githubLookup = await fetchGithubRepoInfo(packageInfo.repositoryUrl);
         }
+        else if (packageName && kind === "uvx-package") {
+            // No PyPI metadata fetcher yet — only cross-check against the known-good
+            // PyPI repo map (never the npm one; see KNOWN_UVX_SERVER_REPOS comment).
+            packageInfo = { name: packageName };
+            const knownRepo = KNOWN_UVX_SERVER_REPOS[packageName];
+            githubLookup = knownRepo ? await fetchGithubRepoInfo(knownRepo) : { status: "not-found" };
+        }
         const { tokens, confidence } = estimateTokens(packageName, kind);
         const category = categorize(serverName, packageName);
         const { score, notes } = computeTrustScore({ githubLookup, kind, packageName });

package/dist/heuristics.js

@@ -9,12 +9,10 @@ const KNOWN_SERVER_TOKEN_COSTS = {
     "@modelcontextprotocol/server-filesystem": 900,
     "@modelcontextprotocol/server-slack": 1200,
     "@modelcontextprotocol/server-postgres": 800,
-    "@modelcontextprotocol/server-sqlite": 700,
     "@modelcontextprotocol/server-puppeteer": 1400,
     "@modelcontextprotocol/server-brave-search": 500,
     "@modelcontextprotocol/server-memory": 400,
     "@modelcontextprotocol/server-everything": 2000,
-    "@modelcontextprotocol/server-fetch": 350,
     "@modelcontextprotocol/server-sequential-thinking": 300,
     "@modelcontextprotocol/server-google-maps": 900,
     "@modelcontextprotocol/server-everart": 500,
@@ -29,8 +27,24 @@ const KNOWN_SERVER_TOKEN_COSTS = {
     "mongodb-mcp-server": 1500,
     "@supabase/mcp-server-supabase": 1400,
     "@cloudflare/mcp-server-cloudflare": 1900,
-    "stripe-mcp": 1700,
-    "@asana/mcp-server": 1200,
+    // NOTE: "stripe-mcp" isn't a real package — Stripe's official npm package is
+    // "@stripe/mcp" (repo: stripe/ai). "@asana/mcp-server" was dropped entirely:
+    // Asana's official MCP is remote-only (OAuth via mcp.asana.com), not something
+    // that shows up as an npx entry in a local config, so a hardcoded cost for it
+    // was never going to match anything real.
+    "@stripe/mcp": 1700,
+};
+/**
+ * Known costs for uvx/pipx (PyPI) packages, kept in a separate table keyed by
+ * the *same string space* as KNOWN_SERVER_TOKEN_COSTS above — but never merged
+ * with it. "mcp-server-fetch" invoked via npx is a different, unrelated package
+ * (an npm-namesquat) from "mcp-server-fetch" invoked via uvx (the real PyPI
+ * tool); estimateTokens() below picks the right table based on `kind` so an
+ * npx-invoked squat can never inherit the trusted PyPI package's numbers.
+ */
+const KNOWN_UVX_TOKEN_COSTS = {
+    "mcp-server-fetch": 350,
+    "mcp-server-sqlite": 700,
 };
 /**
  * Category keywords are matched as whole tokens against the server/package name,
@@ -60,14 +74,19 @@ function tokenize(s) {
         .filter(Boolean));
 }
 export function estimateTokens(packageName, kind) {
-    if (packageName && KNOWN_SERVER_TOKEN_COSTS[packageName] !== undefined) {
-        return { tokens: KNOWN_SERVER_TOKEN_COSTS[packageName], confidence: "known" };
+    if (packageName) {
+        const table = kind === "uvx-package" ? KNOWN_UVX_TOKEN_COSTS : KNOWN_SERVER_TOKEN_COSTS;
+        if (table[packageName] !== undefined) {
+            return { tokens: table[packageName], confidence: "known" };
+        }
     }
     // Heuristic fallback by server "shape"
     if (kind === "remote-http")
         return { tokens: 1500, confidence: "heuristic" };
     if (kind === "local-binary")
         return { tokens: 1000, confidence: "heuristic" };
+    if (kind === "uvx-package")
+        return { tokens: 700, confidence: "heuristic" };
     return { tokens: 800, confidence: "heuristic" };
 }
 export function categorize(name, packageName) {

package/dist/registryInfo.js

@@ -1,5 +1,39 @@
 const NPM_REGISTRY = "https://registry.npmjs.org";
 const GITHUB_API = "https://api.github.com";
+/**
+ * Fallback GitHub repo for packages whose npm registry metadata is missing (or
+ * inconsistent about) a `repository` field. Verified by hand against the live
+ * registry — several official @modelcontextprotocol/* packages simply don't
+ * publish this field even though they live in a real, well-known monorepo, and
+ * without this map their trust score always reads "no repo found" regardless of
+ * how well-maintained the actual project is.
+ */
+const KNOWN_SERVER_REPOS = {
+    "@modelcontextprotocol/server-github": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-slack": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-postgres": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-puppeteer": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-brave-search": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-google-maps": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-everart": "https://github.com/modelcontextprotocol/servers",
+    "@modelcontextprotocol/server-redis": "https://github.com/modelcontextprotocol/servers",
+    "@cloudflare/mcp-server-cloudflare": "https://github.com/cloudflare/mcp-server-cloudflare",
+};
+/**
+ * Repo fallback for known **uvx/pipx** (PyPI) packages only. Deliberately kept
+ * separate from KNOWN_SERVER_REPOS above: "mcp-server-fetch" and
+ * "mcp-server-sqlite" are real PyPI package names, but those exact strings are
+ * ALSO squattable (or already squatted — see e.g. the "npx confusion" security
+ * research canary published to npm under the literal name "mcp-server-fetch").
+ * If a config invokes a same-named package via `npx` instead of `uvx`, it is a
+ * *different* package in a *different* registry, and must never inherit this
+ * trust/repo mapping. Kind separation in classifyServer() is what keeps these
+ * two lookups from colliding.
+ */
+export const KNOWN_UVX_SERVER_REPOS = {
+    "mcp-server-fetch": "https://github.com/modelcontextprotocol/servers",
+    "mcp-server-sqlite": "https://github.com/modelcontextprotocol/servers",
+};
 /** Figures out roughly what kind of server this is, and pulls out an npm package name if present. */
 export function classifyServer(cfg) {
     if (cfg.url) {
@@ -16,9 +50,13 @@ export function classifyServer(cfg) {
         }
     }
     if (command === "uvx" || command === "pipx") {
+        // Separate kind from npx: uvx/pipx pull from PyPI, not npm. Same package-name
+        // string can refer to two completely unrelated packages across the two
+        // registries (see the KNOWN_UVX_SERVER_REPOS comment), so these must never
+        // be looked up against the npm registry as if they were npx packages.
         const pkg = args.find((a) => !a.startsWith("-"));
         if (pkg) {
-            return { kind: "npx-package", packageName: stripVersion(pkg), rawCommandLine };
+            return { kind: "uvx-package", packageName: stripVersion(pkg), rawCommandLine };
         }
     }
     if (command) {
@@ -53,7 +91,8 @@ export async function fetchNpmPackageInfo(packageName) {
         const latestVersion = data["dist-tags"]?.latest;
         const versionData = latestVersion ? data.versions?.[latestVersion] : undefined;
         const repository = versionData?.repository ?? data.repository;
-        const repoUrl = typeof repository === "string" ? repository : repository?.url;
+        const repoUrl = (typeof repository === "string" ? repository : repository?.url) ??
+            KNOWN_SERVER_REPOS[packageName];
         return {
             name: packageName,
             version: latestVersion,

package/package.json

@@ -1,8 +1,18 @@
 {
   "name": "mcp-tollbooth",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "How much context are your MCP servers actually costing you? Token budget + duplicate-server detection for Claude Desktop, Claude Code, Cursor and VS Code.",
-  "keywords": ["mcp", "model-context-protocol", "claude", "claude-code", "cursor", "ai-agents", "context-window", "token-budget", "cli"],
+  "keywords": [
+    "mcp",
+    "model-context-protocol",
+    "claude",
+    "claude-code",
+    "cursor",
+    "ai-agents",
+    "context-window",
+    "token-budget",
+    "cli"
+  ],
   "license": "MIT",
   "type": "module",
   "author": "Ereb-Blade",