diff --git a/packages/di/src/tokenUniqueness.test.ts b/packages/di/src/tokenUniqueness.test.ts
new file mode 100644
index 000000000..79274200b
--- /dev/null
+++ b/packages/di/src/tokenUniqueness.test.ts
@@ -0,0 +1,102 @@
+import { readdirSync, readFileSync } from "node:fs";
+import { join } from "node:path";
+import { fileURLToPath } from "node:url";
+import { describe, expect, it } from "vitest";
+
+// Guards against the Symbol.for collision class: `Symbol.for(key)` resolves
+// through the global registry, so two token consts in different files (even
+// different packages, with different export names) that pass the SAME string
+// are the SAME symbol. Binding two different services to it is last-load-wins
+// with no error — the wrong service resolves (e.g. "r.connect is not a
+// function"). Every DI token string must therefore be defined in exactly one
+// production source file.
+
+const repoRoot = fileURLToPath(new URL("../../..", import.meta.url));
+const SCAN_ROOTS = ["packages", "apps"];
+const SKIP_DIRS = new Set([
+  "node_modules",
+  "dist",
+  "build",
+  "out",
+  ".vite",
+  ".turbo",
+  "coverage",
+]);
+const SOURCE_RE = /\.tsx?$/;
+// Tests legitimately re-declare a real token's string to match it in a fake
+// container; stories and type decls are not bindings.
+const EXCLUDE_FILE_RE = /\.(test|spec|stories)\.tsx?$|\.d\.ts$/;
+const SYMBOL_FOR_RE = /Symbol\.for\(\s*["']([^"']+)["']/g;
+const NAMED_TOKEN_RE =
+  /const\s+([A-Z][A-Z0-9_]*)\s*=\s*Symbol\.for\(\s*["']([^"']+)["']/g;
+
+function collectSourceFiles(dir: string, acc: string[]): void {
+  for (const entry of readdirSync(dir, { withFileTypes: true })) {
+    if (entry.isDirectory()) {
+      if (SKIP_DIRS.has(entry.name)) continue;
+      collectSourceFiles(join(dir, entry.name), acc);
+    } else if (
+      SOURCE_RE.test(entry.name) &&
+      !EXCLUDE_FILE_RE.test(entry.name)
+    ) {
+      acc.push(join(dir, entry.name));
+    }
+  }
+}
+
+interface TokenSite {
+  file: string;
+  name: string;
+}
+
+function findTokenStrings(): Map<string, TokenSite[]> {
+  const files: string[] = [];
+  for (const root of SCAN_ROOTS) {
+    collectSourceFiles(join(repoRoot, root), files);
+  }
+
+  const byString = new Map<string, TokenSite[]>();
+  for (const file of files) {
+    const text = readFileSync(file, "utf8");
+    if (!text.includes("Symbol.for(")) continue;
+
+    const names = new Map<string, string>();
+    for (const match of text.matchAll(NAMED_TOKEN_RE)) {
+      names.set(match[2], match[1]);
+    }
+
+    const relative = file.slice(repoRoot.length);
+    for (const match of text.matchAll(SYMBOL_FOR_RE)) {
+      const key = match[1];
+      const sites = byString.get(key) ?? [];
+      sites.push({ file: relative, name: names.get(key) ?? "(unnamed)" });
+      byString.set(key, sites);
+    }
+  }
+  return byString;
+}
+
+describe("DI token Symbol.for uniqueness", () => {
+  it("never defines the same Symbol.for string in more than one source file", () => {
+    const byString = findTokenStrings();
+
+    const collisions = [...byString.entries()].filter(([, sites]) => {
+      const distinctFiles = new Set(sites.map((s) => s.file));
+      return distinctFiles.size > 1;
+    });
+
+    const report = collisions
+      .map(([key, sites]) => {
+        const lines = sites.map((s) => `    ${s.name} @ ${s.file}`).join("\n");
+        return `  Symbol.for("${key}") is defined in multiple files:\n${lines}`;
+      })
+      .join("\n\n");
+
+    expect(
+      collisions.length,
+      collisions.length === 0
+        ? ""
+        : `Duplicate DI token string(s) found — these resolve to the SAME symbol and will silently shadow each other. Namespace them distinctly (posthog.<area>.<thing>):\n\n${report}\n`,
+    ).toBe(0);
+  });
+});