fix: redact the code search query before it leaves the machine

src/main/search/code_index.test.ts

@@ -3,6 +3,7 @@ import { tmpdir } from 'node:os';
 import { join } from 'node:path';
 import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest';
 import { SettingsStore, type SecretCipher } from '../settings';
+import { REDACTION_MARK } from '../tools/redact';
 import { ShellRunner } from '../tools/shell';
 import { Workspace } from '../tools/workspace';
 import { chunkFile } from './chunker';
@@ -485,6 +486,21 @@ describe('CodeIndex with a reranker', () => {
     expect(documents).not.toContain('hunter2hunter2');
   });
 
+  it('masks secrets in the query before it leaves to the embedding and the reranker (#286)', async () => {
+    const embedder = new FakeEmbedder();
+    const reranker = new FakeReranker('addToCart');
+    const index = new CodeIndex(new Workspace(root), embedder, indexDir, () => 1000, reranker);
+    const token = `ghp_${'a'.repeat(36)}`;
+
+    // A model can copy a token it read straight into its search query, and both services are third parties.
+    await index.searchDetailed(`fix the login that stores ${token}`, 2, signal);
+
+    expect(embedder.inputs.filter((input) => input === 'query')).toHaveLength(1);
+    const sent = embedder.calls[embedder.inputs.indexOf('query')];
+    expect(sent).toEqual([`fix the login that stores ${REDACTION_MARK}`]);
+    expect(reranker.seen!.query).toBe(`fix the login that stores ${REDACTION_MARK}`);
+  });
+
   it('falls back to the embedding order when reranking fails, and says so', async () => {
     const reranker = new FakeReranker('addToCart');
     reranker.fail = new Error('OpenRouter rerank request failed (503)');

src/main/search/code_index.ts

@@ -196,7 +196,10 @@ export class CodeIndex implements CodeSearch {
     onProgress?: (progress: UpdateProgress) => void,
   ): Promise<SearchResult> {
     await this.update(signal, onProgress);
-    const [queryVector] = await this.embedder.embed([query], 'query', signal);
+    // The query leaves the machine for the embedding and the rerank service, so a secret the model copied into it
+    // is masked first, the way a web search query is (#73) and the way the documents below already are.
+    const sent = redactSecrets(query);
+    const [queryVector] = await this.embedder.embed([sent], 'query', signal);
     if (!queryVector) throw new Error('The embedding service returned no vector for the query.');
     const q = normalize(Float32Array.from(queryVector));
 
@@ -231,7 +234,7 @@ export class CodeIndex implements CodeSearch {
 
     try {
       const ranked = await this.reranker.rerank(
-        query,
+        sent,
         candidates.map((hit) => redactSecrets(`${hit.path}\n${hit.text}`)),
         Math.min(limit, candidates.length),
         signal,