Propose a patch

Evidence

The sandbox observed nothing beyond the static findings; the proposal cites those.

diff --git a/src/xml.js b/src/xml.js
--- a/src/xml.js
+++ b/src/xml.js
@@ -25,1 +25,0 @@
-const roots = normalise(parser.parse(text.replace(/^/, '')));
diff --git a/src/song.js b/src/song.js
--- a/src/song.js
+++ b/src/song.js
@@ -248,1 +248,0 @@
-const body = text.replace(/^/, '').trimStart();
@@ -393,1 +392,0 @@
-notes: (zip.text('notes.txt') || '').replace(/^/, ''),
diff --git a/src/plugins.js b/src/plugins.js
--- a/src/plugins.js
+++ b/src/plugins.js
@@ -36,1 +36,0 @@
-const text = (n) => (entries[n] ? strFromU8(entries[n]).replace(/^/, '') : null);
diff --git a/test/edittask.test.js b/test/edittask.test.js
--- a/test/edittask.test.js
+++ b/test/edittask.test.js
@@ -20,1 +20,0 @@
-createTextFile: (u) => { let buf = ''; return { writeString: (s) => (buf += s), close: () => files.set(u.url, '' + buf) }; },
@@ -31,1 +30,0 @@
-result: () => JSON.parse(files.get(MAILBOX + 'edit-result.json').replace(/^/, '')),
diff --git a/device/StudioOneMCP/BridgeCore.js b/device/StudioOneMCP/BridgeCore.js
--- a/device/StudioOneMCP/BridgeCore.js
+++ b/device/StudioOneMCP/BridgeCore.js
@@ -60,1 +60,0 @@
-return typeof text === "string" && text ? JSON.parse(text.replace(/^/, "")) : null;
diff --git a/device/EditTasks/package/McpEdit.js b/device/EditTasks/package/McpEdit.js
--- a/device/EditTasks/package/McpEdit.js
+++ b/device/EditTasks/package/McpEdit.js
@@ -26,1 +26,0 @@
-try { value = JSON.parse(text.replace(/^/, "")); } catch (_) { value = null; }
diff --git a/test/core.test.js b/test/core.test.js
--- a/test/core.test.js
+++ b/test/core.test.js
@@ -31,1 +31,0 @@
-assert.ok(host.client.raw('status.json').startsWith(''));
diff --git a/test/helpers/fixtures.js b/test/helpers/fixtures.js
--- a/test/helpers/fixtures.js
+++ b/test/helpers/fixtures.js
@@ -10,1 +10,0 @@
-const songXml = `<?xml version="1.0" encoding="UTF-8"?>
@@ -134,1 +133,0 @@
-files['Presets/Channels/Vox/1 - Pro EQ.fxpreset'] = strToU8('<AudioEffectPreset><Attributes x:id="ParameterData" lffreq="40" lfgain="-3.5"/></AudioEffectPreset>');
@@ -138,1 +136,0 @@
-files['notes.txt'] = strToU8('Verse 1: keep the breath before the chorus');
@@ -139,1 +136,0 @@
-files['notepad.xml'] = strToU8(`<?xml version="1.0" encoding="UTF-8"?>
diff --git a/test/server.test.js b/test/server.test.js
--- a/test/server.test.js
+++ b/test/server.test.js
@@ -135,1 +135,0 @@
-writeFileSync(join(mcpHome, 'mailbox', 'status.json'), '' + JSON.stringify({ protocol: 1, closed: true }));
diff --git a/test/plugins.test.js b/test/plugins.test.js
--- a/test/plugins.test.js
+++ b/test/plugins.test.js
@@ -10,1 +10,0 @@
-const meta = (cls, dataFile) => `<?xml version="1.0" encoding="UTF-8"?>
diff --git a/test/helpers/s1host.js b/test/helpers/s1host.js
--- a/test/helpers/s1host.js
+++ b/test/helpers/s1host.js
@@ -132,1 +132,0 @@
-return { writeString: (s) => (buf += s), close: () => files.set(u.url, '' + buf) };
@@ -164,1 +163,0 @@
-return t === undefined ? null : JSON.parse(t.replace(/^/, ''));

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.4.0. If the behavior is intended, declare it (host under permissions.network, a line in the README) instead of removing it; the next crawl scores declarations.

Issue / PR body

Title: fender-studio-mcp: Hidden zero-width characters detected (KillerSkills verdict caution 73)

## What the honeypot observed
KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.7) and scored it **CAUTION 73/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/glama/fender-studio-mcp/0.4.0

## Findings with a location
- `src/xml.js:25` · Hidden zero-width characters detected — src/xml.js:25 · Hidden zero-width characters detected — may conceal content
- `src/song.js:393` · Hidden zero-width characters detected — src/song.js:393 · Hidden zero-width characters detected — may conceal content
- `src/plugins.js:36` · Hidden zero-width characters detected — src/plugins.js:36 · Hidden zero-width characters detected — may conceal content
- `test/edittask.test.js:20` · Hidden zero-width characters detected — test/edittask.test.js:20 · Hidden zero-width characters detected — may conceal content
- `device/StudioOneMCP/BridgeCore.js:60` · Hidden zero-width characters detected — device/StudioOneMCP/BridgeCore.js:60 · Hidden zero-width characters detected — may conceal content
- `src/song.js:248` · Hidden zero-width characters detected — src/song.js:248 · Hidden zero-width characters detected — may conceal content
- `device/EditTasks/package/McpEdit.js:26` · Hidden zero-width characters detected — device/EditTasks/package/McpEdit.js:26 · Hidden zero-width characters detected — may conceal content
- `test/core.test.js:31` · Hidden zero-width characters detected — test/core.test.js:31 · Hidden zero-width characters detected — may conceal content
- `test/helpers/fixtures.js:139` · Hidden zero-width characters detected — test/helpers/fixtures.js:139 · Hidden zero-width characters detected — may conceal content
- `test/helpers/fixtures.js:10` · Hidden zero-width characters detected — test/helpers/fixtures.js:10 · Hidden zero-width characters detected — may conceal content
- `test/helpers/fixtures.js:138` · Hidden zero-width characters detected — test/helpers/fixtures.js:138 · Hidden zero-width characters detected — may conceal content
- `test/edittask.test.js:31` · Hidden zero-width characters detected — test/edittask.test.js:31 · Hidden zero-width characters detected — may conceal content

## Proposed change
The attached patch removes the 17 lines above. If the behavior is intended, the fix is to declare it instead: list the host under `permissions.network` (or the MCP manifest) and say in the README why the package needs it; the next crawl will re-score it with the declaration taken into account.

## How this will be verified
Nothing is "verified safe" by assertion. When a new version is published, KillerSkills re-runs it in the honeypot and the verdict page updates; you can trigger it early from the verdict page ("Re-scan after I publish a fix"). Reply here or at [email protected] if you believe the verdict is wrong; disputes are reviewed by a person and the outcome is published on the verdict page.

_Proposed through killerskills.ai/threats by a user of the package, under their own GitHub account. The evidence is public; the package was never modified by KillerSkills._