Memory: revalidate multimodal files before indexing

src/memory/internal.test.ts

@@ -214,6 +214,28 @@ describe("buildFileEntry", () => {
     ]);
     expect(built?.structuredInputBytes).toBeGreaterThan(0);
   });
+
+  it("skips lazy multimodal indexing when the file grows after discovery", async () => {
+    const tmpDir = getTmpDir();
+    const target = path.join(tmpDir, "diagram.png");
+    await fs.writeFile(target, Buffer.from("png"));
+
+    const entry = await buildFileEntry(target, tmpDir, multimodal);
+    await fs.writeFile(target, Buffer.alloc(entry!.size + 32, 1));
+
+    await expect(buildMultimodalChunkForIndexing(entry!)).resolves.toBeNull();
+  });
+
+  it("skips lazy multimodal indexing when file bytes change after discovery", async () => {
+    const tmpDir = getTmpDir();
+    const target = path.join(tmpDir, "diagram.png");
+    await fs.writeFile(target, Buffer.from("png"));
+
+    const entry = await buildFileEntry(target, tmpDir, multimodal);
+    await fs.writeFile(target, Buffer.from("gif"));
+
+    await expect(buildMultimodalChunkForIndexing(entry!)).resolves.toBeNull();
+  });
 });
 
 describe("chunkMarkdown", () => {

src/memory/internal.ts

@@ -20,6 +20,7 @@ export type MemoryFileEntry = {
   mtimeMs: number;
   size: number;
   hash: string;
+  dataHash?: string;
   kind?: "markdown" | "multimodal";
   contentText?: string;
   modality?: MemoryMultimodalModality;
@@ -234,6 +235,7 @@ export async function buildFileEntry(
       mtimeMs: stat.mtimeMs,
       size: stat.size,
       hash: chunkHash,
+      dataHash,
       kind: "multimodal",
       contentText,
       modality,
@@ -261,11 +263,26 @@ export async function buildFileEntry(
 }
 
 async function loadMultimodalEmbeddingInput(
-  entry: Pick<MemoryFileEntry, "absPath" | "contentText" | "mimeType" | "kind">,
+  entry: Pick<
+    MemoryFileEntry,
+    "absPath" | "contentText" | "mimeType" | "kind" | "size" | "dataHash"
+  >,
 ): Promise<EmbeddingInput | null> {
   if (entry.kind !== "multimodal" || !entry.contentText || !entry.mimeType) {
     return null;
   }
+  let stat;
+  try {
+    stat = await fs.stat(entry.absPath);
+  } catch (err) {
+    if (isFileMissingError(err)) {
+      return null;
+    }
+    throw err;
+  }
+  if (stat.size !== entry.size) {
+    return null;
+  }
   let buffer: Buffer;
   try {
     buffer = await fs.readFile(entry.absPath);
@@ -275,6 +292,10 @@ async function loadMultimodalEmbeddingInput(
     }
     throw err;
   }
+  const dataHash = crypto.createHash("sha256").update(buffer).digest("hex");
+  if (entry.dataHash && entry.dataHash !== dataHash) {
+    return null;
+  }
   return {
     text: entry.contentText,
     parts: [
@@ -289,7 +310,10 @@ async function loadMultimodalEmbeddingInput(
 }
 
 export async function buildMultimodalChunkForIndexing(
-  entry: Pick<MemoryFileEntry, "absPath" | "contentText" | "mimeType" | "kind" | "hash">,
+  entry: Pick<
+    MemoryFileEntry,
+    "absPath" | "contentText" | "mimeType" | "kind" | "hash" | "size" | "dataHash"
+  >,
 ): Promise<MultimodalMemoryChunk | null> {
   const embeddingInput = await loadMultimodalEmbeddingInput(entry);
   if (!embeddingInput) {