harper-knowledge 0.1.2 → 0.2.0

package/dist/core/embeddings.js

@@ -152,9 +152,10 @@ export async function initEmbeddingModel(config) {
         return;
     }
     const modelPath = await downloadModelIfNeeded(modelName);
-    // Load the model with node-llama-cpp
+    // Load the model with node-llama-cpp (use prebuilt binaries only —
+    // build: "never" avoids requiring cmake/make on deployment targets)
     const { getLlama } = (await import("node-llama-cpp"));
-    llama = await getLlama({ progressLogs: false });
+    llama = await getLlama({ progressLogs: false, build: "never" });
     embeddingModel = await llama.loadModel({ modelPath });
     embeddingContext = await embeddingModel.createEmbeddingContext({
         contextSize: "auto",

package/dist/core/entries.d.ts

@@ -72,6 +72,19 @@ export declare function linkSupersedes(newId: string, oldId: string): Promise<vo
  * @throws Error if any entry does not exist
  */
 export declare function linkSiblings(ids: string[]): Promise<void>;
+/**
+ * Reindex embeddings for all entries missing them.
+ *
+ * Iterates every KnowledgeEntry, generates embeddings for any that
+ * don't have one yet, and writes them back. Returns counts of
+ * processed, updated, and failed entries.
+ */
+export declare function reindexEmbeddings(): Promise<{
+    total: number;
+    updated: number;
+    failed: number;
+    skipped: number;
+}>;
 /**
  * Link two entries as related.
  *

package/dist/core/entries.js

@@ -209,6 +209,44 @@ export async function linkSiblings(ids) {
         });
     }
 }
+/**
+ * Reindex embeddings for all entries missing them.
+ *
+ * Iterates every KnowledgeEntry, generates embeddings for any that
+ * don't have one yet, and writes them back. Returns counts of
+ * processed, updated, and failed entries.
+ */
+export async function reindexEmbeddings() {
+    let total = 0;
+    let updated = 0;
+    let failed = 0;
+    let skipped = 0;
+    for await (const record of databases.kb.KnowledgeEntry.search({})) {
+        total++;
+        const entry = record;
+        // Skip entries that already have embeddings
+        if (entry.embedding && entry.embedding.length > 0) {
+            skipped++;
+            continue;
+        }
+        try {
+            const embeddingText = `${entry.title}\n\n${entry.content}`;
+            const embedding = await generateEmbedding(embeddingText);
+            await databases.kb.KnowledgeEntry.put({
+                ...record,
+                id: entry.id,
+                embedding,
+            });
+            updated++;
+            logger?.info?.(`Reindexed embedding for entry: ${entry.id}`);
+        }
+        catch (error) {
+            failed++;
+            logger?.warn?.(`Failed to reindex entry ${entry.id}:`, error.message);
+        }
+    }
+    return { total, updated, failed, skipped };
+}
 /**
  * Link two entries as related.
  *

package/dist/mcp/tools.js

@@ -6,7 +6,7 @@
  */
 import * as z from "zod/v4";
 import { search } from "../core/search.js";
-import { createEntry, getEntry, updateEntry, stripEmbedding, } from "../core/entries.js";
+import { createEntry, getEntry, updateEntry, stripEmbedding, reindexEmbeddings, } from "../core/entries.js";
 import { listTags } from "../core/tags.js";
 import { submitTriage } from "../core/triage.js";
 import { generateEmbedding } from "../core/embeddings.js";
@@ -494,4 +494,29 @@ export function registerTools(server, caller) {
             return errorContent("Failed to get edit history. Please try again.");
         }
     });
+    // =========================================================================
+    // 9. knowledge_reindex — Backfill missing embeddings
+    // =========================================================================
+    server.registerTool("knowledge_reindex", {
+        description: "Reindex embeddings for all knowledge entries that are missing them. " +
+            "Use this after the embedding model becomes available on a deployment " +
+            "where entries were initially created without embeddings. " +
+            "Requires write access.",
+        inputSchema: {},
+    }, async (_input, { authInfo }) => {
+        const caller = authInfo;
+        if (!caller?.scopes?.includes("mcp:write")) {
+            return errorContent("Write access required to reindex embeddings.");
+        }
+        try {
+            const result = await reindexEmbeddings();
+            return jsonContent({
+                message: "Embedding reindex complete",
+                ...result,
+            });
+        }
+        catch (error) {
+            return errorContent("Failed to reindex embeddings. Please try again.");
+        }
+    });
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "harper-knowledge",
-  "version": "0.1.2",
+  "version": "0.2.0",
   "description": "Knowledge base plugin for Harper with MCP server integration",
   "type": "module",
   "main": "dist/index.js",
@@ -55,6 +55,9 @@
     "prettier": "3.8.1",
     "typescript": "5.9.3"
   },
+  "optionalDependencies": {
+    "@node-llama-cpp/linux-x64": "^3.15.1"
+  },
   "peerDependencies": {
     "harperdb": ">=4.7.0"
   },