fix(llm-security): normalization/discovery evasion + SIG embedded-base64 & custom rules (#21,#23,#30,#36,#42,#52,#55)

#21 the bash normalizer decoded only \xHH, leaving ANSI-C octal/\u/\U forms literal so canonical rm/curl never surfaced; now decodes all three. #23 file-discovery keyed on extname so .env.local/.env.example (extname .local/.example) were silently skipped; now matches multi-part suffixes. #42 a legitimate leading UTF-8 BOM was flagged HIGH (and the tool's own auto-cleaner refused to strip it); pos-0 BOM now excepted. #52 collapseLetterSpacing used a literal space, letting multi-space/tab spacing evade; now [ \t]+. #55 redact(_,60,0) did slice(-0) and leaked the whole unredacted URL; showEnd===0 now means no tail.

#30 embedded base64 (const x = "<base64>") never satisfied the whole-string decode, so the SIG identity engine never saw it; added decodeEmbeddedBase64 as an OPT-IN param on normalizeForScan (default off — appending a decoded copy would double-count per-match findings, e.g. content-extractor's injection scan) and enabled it only in signature-scanner, which dedups variants. #36 signature-scanner ignored the documented sig.custom_rules_path policy option; now loads+merges custom rules through the same family filter. Suite 2004/0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Claude-Session: https://claude.ai/code/session_01TcQyMTQfyrsAapaCMPxTtQ

tests/scanners/unicode-bom.test.mjs

@@ -0,0 +1,97 @@
+// unicode-bom.test.mjs — Regression tests for the leading-BOM exception in
+// unicode-scanner (v7.8.3, #42).
+//
+// A legitimate UTF-8 BOM (U+FEFF at file position 0) was flagged as a HIGH
+// zero-width finding — a finding the tool's own auto-cleaner refuses to strip
+// (auto-cleaner.mjs preserves cp === 0xFEFF at line 0, pos 0). The scanner
+// must apply the same file-start exception. A BOM anywhere else in the file
+// is still a zero-width smuggling vector and must be flagged.
+
+import { describe, it, beforeEach, after } from 'node:test';
+import assert from 'node:assert/strict';
+import { join } from 'node:path';
+import { tmpdir } from 'node:os';
+import { mkdtemp, writeFile, rm } from 'node:fs/promises';
+import { resetCounter } from '../../scanners/lib/output.mjs';
+import { discoverFiles } from '../../scanners/lib/file-discovery.mjs';
+import { scan } from '../../scanners/unicode-scanner.mjs';
+
+const created = [];
+
+/** Make a scan root containing a single file with the given content. */
+async function makeRepo(content, fileName = 'app.mjs') {
+  const root = await mkdtemp(join(tmpdir(), 'llmsec-bom-'));
+  created.push(root);
+  await writeFile(join(root, fileName), content, 'utf8');
+  return root;
+}
+
+async function scanRepo(root) {
+  resetCounter();
+  const discovery = await discoverFiles(root);
+  return scan(root, discovery);
+}
+
+function zeroWidthFindings(result) {
+  return result.findings.filter(f =>
+    f.title.toLowerCase().includes('zero-width') ||
+    (f.evidence && f.evidence.toUpperCase().includes('U+FEFF'))
+  );
+}
+
+after(async () => {
+  for (const dir of created) await rm(dir, { recursive: true, force: true });
+});
+
+describe('unicode-scanner — leading UTF-8 BOM exception (v7.8.3, #42)', () => {
+  beforeEach(() => {
+    resetCounter();
+  });
+
+  it('does not flag a legitimate BOM at file position 0', async () => {
+    const root = await makeRepo('\uFEFFconst x = 1;\nexport default x;\n');
+    const result = await scanRepo(root);
+    assert.equal(result.status, 'ok');
+    const hits = zeroWidthFindings(result);
+    assert.equal(
+      hits.length, 0,
+      `leading BOM must not be flagged, got: ${hits.map(f => f.title).join('; ')}`
+    );
+  });
+
+  it('still flags a BOM at the start of a later line', async () => {
+    const root = await makeRepo('const x = 1;\n\uFEFFconst y = 2;\n');
+    const result = await scanRepo(root);
+    assert.equal(result.status, 'ok');
+    const hits = zeroWidthFindings(result);
+    assert.ok(
+      hits.length >= 1,
+      `mid-file BOM must be flagged, got ${hits.length} findings`
+    );
+    assert.equal(hits[0].line, 2);
+  });
+
+  it('still flags a BOM mid-line on the first line', async () => {
+    const root = await makeRepo('const x\uFEFF = 1;\n');
+    const result = await scanRepo(root);
+    assert.equal(result.status, 'ok');
+    const hits = zeroWidthFindings(result);
+    assert.ok(
+      hits.length >= 1,
+      `first-line non-leading BOM must be flagged, got ${hits.length} findings`
+    );
+  });
+
+  it('still flags other zero-width chars on the first line after a BOM', async () => {
+    const root = await makeRepo('\uFEFFconst x\u200B = 1;\n');
+    const result = await scanRepo(root);
+    assert.equal(result.status, 'ok');
+    const hits = result.findings.filter(f =>
+      f.evidence && f.evidence.toUpperCase().includes('U+200B')
+    );
+    assert.ok(
+      hits.length >= 1,
+      `U+200B after a leading BOM must still be flagged, got ${hits.length} findings`
+    );
+  });
+});