From 030ed5d5926c038b0e54c3afb0e7c92e47f7c695 Mon Sep 17 00:00:00 2001
From: Peter Steinberger <steipete@gmail.com>
Date: Sat, 17 Jan 2026 21:58:59 +0000
Subject: [PATCH] fix: skip empty memory chunks

---
 CHANGELOG.md                                 |  1 +
 src/memory/manager.embedding-batches.test.ts | 29 ++++++++++++++++++++
 src/memory/manager.ts                        |  4 ++-
 3 files changed, 33 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 2f5ecd4b6..c1466b408 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,6 +10,7 @@ Docs: https://docs.clawd.bot
 - Tools: show exec elevated flag before the command and keep it outside markdown in tool summaries.
 - Memory: parallelize embedding indexing with rate-limit retries.
 - Memory: split overly long lines to keep embeddings under token limits.
+- Memory: skip empty chunks to avoid invalid embedding inputs.
 
 ## 2026.1.17-1
 
diff --git a/src/memory/manager.embedding-batches.test.ts b/src/memory/manager.embedding-batches.test.ts
index a11839db2..6a0b7e505 100644
--- a/src/memory/manager.embedding-batches.test.ts
+++ b/src/memory/manager.embedding-batches.test.ts
@@ -191,4 +191,33 @@ describe("memory embedding batches", () => {
 
     expect(calls).toBe(3);
   }, 10000);
+
+  it("skips empty chunks so embeddings input stays valid", async () => {
+    await fs.writeFile(path.join(workspaceDir, "memory", "2026-01-07.md"), "\n\n\n");
+
+    const cfg = {
+      agents: {
+        defaults: {
+          workspace: workspaceDir,
+          memorySearch: {
+            provider: "openai",
+            model: "mock-embed",
+            store: { path: indexPath },
+            sync: { watch: false, onSessionStart: false, onSearch: false },
+            query: { minScore: 0 },
+          },
+        },
+        list: [{ id: "main", default: true }],
+      },
+    };
+
+    const result = await getMemorySearchManager({ cfg, agentId: "main" });
+    expect(result.manager).not.toBeNull();
+    if (!result.manager) throw new Error("manager missing");
+    manager = result.manager;
+    await manager.sync({ force: true });
+
+    const inputs = embedBatch.mock.calls.flatMap((call) => call[0] ?? []);
+    expect(inputs).not.toContain("");
+  });
 });
diff --git a/src/memory/manager.ts b/src/memory/manager.ts
index b5d638fa0..58afa0250 100644
--- a/src/memory/manager.ts
+++ b/src/memory/manager.ts
@@ -1091,7 +1091,9 @@ export class MemoryIndexManager {
     options: { source: MemorySource; content?: string },
   ) {
     const content = options.content ?? (await fs.readFile(entry.absPath, "utf-8"));
-    const chunks = chunkMarkdown(content, this.settings.chunking);
+    const chunks = chunkMarkdown(content, this.settings.chunking).filter(
+      (chunk) => chunk.text.trim().length > 0,
+    );
     const embeddings = await this.embedChunksInBatches(chunks);
     const sample = embeddings.find((embedding) => embedding.length > 0);
     const vectorReady = sample ? await this.ensureVectorReady(sample.length) : false;