fix(media): distrust image hints for container bytes

CHANGELOG.md

@@ -17,6 +17,7 @@ Docs: https://docs.openclaw.ai
 ### Fixes
 
 - MCP plugin tools: forward host MCP `tools/call` `AbortSignal` through `createPluginToolsMcpHandlers().callTool` into plugin `tool.execute`, so host cancellation actually cancels in-flight plugin tool calls instead of letting them run to completion. (#82443) Thanks @joshavant.
+- Media: ignore image MIME and filename hints when bytes sniff as generic containers, so zip/octet-stream payloads mislabeled as images do not become local image media.
 - Update/doctor: avoid materializing `groupAllowFrom` for channel schemas that reject it, so package-swap doctor repairs do not fail on externalized Slack configs.
 - Gateway/media: prevent image filenames from overriding generic non-image byte sniffing, so zip/octet-stream payloads mislabeled as images are offloaded or rejected before they become inline image attachments.
 - Plugins/web search: downgrade stale optional provider installs to warnings so Gateway and doctor repair paths keep running after startup provider selection. Refs #82313. Thanks @crackmac.

src/media/mime.test.ts

@@ -78,6 +78,30 @@ describe("mime detection", () => {
       },
       expected: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
     },
+    {
+      name: "does not let image extensions override generic zip bytes",
+      input: async () => {
+        const zip = new JSZip();
+        zip.file("hello.txt", "hi");
+        return {
+          buffer: await zip.generateAsync({ type: "nodebuffer" }),
+          filePath: "/tmp/fake.png",
+        };
+      },
+      expected: "application/zip",
+    },
+    {
+      name: "does not let image headers override generic zip bytes",
+      input: async () => {
+        const zip = new JSZip();
+        zip.file("hello.txt", "hi");
+        return {
+          buffer: await zip.generateAsync({ type: "nodebuffer" }),
+          headerMime: "image/png",
+        };
+      },
+      expected: "application/zip",
+    },
     {
       name: "uses extension mapping for JavaScript assets",
       input: async () => ({

src/media/mime.ts

@@ -190,6 +190,10 @@ function isGenericMime(mime?: string): boolean {
   return m === "application/octet-stream" || m === "application/zip";
 }
 
+function isImageMime(mime?: string): boolean {
+  return mediaKindFromMime(normalizeMimeType(mime)) === "image";
+}
+
 async function detectMimeImpl(opts: {
   buffer?: Buffer;
   headerMime?: string | null;
@@ -200,23 +204,27 @@ async function detectMimeImpl(opts: {
 
   const headerMime = normalizeMimeType(opts.headerMime);
   const sniffed = await sniffMime(opts.buffer);
+  const sniffedGenericContainer = sniffed && isGenericMime(sniffed);
+  const trustedExtMime = sniffedGenericContainer && isImageMime(extMime) ? undefined : extMime;
+  const trustedHeaderMime =
+    sniffedGenericContainer && isImageMime(headerMime) ? undefined : headerMime;
 
   // Prefer sniffed types, but don't let generic container types override a more
   // specific extension mapping (e.g. XLSX vs ZIP).
-  if (sniffed && (!isGenericMime(sniffed) || !extMime)) {
+  if (sniffed && (!isGenericMime(sniffed) || !trustedExtMime)) {
     return sniffed;
   }
-  if (extMime) {
-    return extMime;
+  if (trustedExtMime) {
+    return trustedExtMime;
   }
-  if (headerMime && !isGenericMime(headerMime)) {
-    return headerMime;
+  if (trustedHeaderMime && !isGenericMime(trustedHeaderMime)) {
+    return trustedHeaderMime;
   }
   if (sniffed) {
     return sniffed;
   }
-  if (headerMime) {
-    return headerMime;
+  if (trustedHeaderMime) {
+    return trustedHeaderMime;
   }
 
   return undefined;

src/media/web-media.test.ts

@@ -1,6 +1,7 @@
 import fs from "node:fs/promises";
 import path from "node:path";
 import { pathToFileURL } from "node:url";
+import JSZip from "jszip";
 import { afterAll, beforeAll, describe, expect, it, vi } from "vitest";
 import { resolveStateDir } from "../config/paths.js";
 import { resolvePreferredOpenClawTmpDir } from "../infra/tmp-openclaw-dir.js";
@@ -308,6 +309,19 @@ describe("loadWebMedia", () => {
     expect(result.buffer.length).toBeGreaterThan(0);
   });
 
+  it("does not treat image-named generic container bytes as local image media", async () => {
+    const zip = new JSZip();
+    zip.file("hello.txt", "hi");
+    const fakeImage = path.join(fixtureRoot, "fake.png");
+    await fs.writeFile(fakeImage, await zip.generateAsync({ type: "nodebuffer" }));
+
+    const result = await loadWebMedia(fakeImage, createLocalWebMediaOptions());
+
+    expect(result.kind).toBe("document");
+    expect(result.contentType).toBe("application/zip");
+    expect(result.fileName).toBe("fake.png");
+  });
+
   it("uses only the leaf filename from Windows-style sandbox-validated media paths", async () => {
     const result = await loadWebMedia(String.raw`C:\workspace\captures\tiny.png`, {
       maxBytes: 1024 * 1024,