Propose a patch

Evidence

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

diff --git a/dist/meta-proxy.d.ts b/dist/meta-proxy.d.ts
--- a/dist/meta-proxy.d.ts
+++ b/dist/meta-proxy.d.ts
@@ -21,1 +21,0 @@
-export declare function metaProxyTokenPath(home?: string): string;
@@ -25,1 +24,0 @@
-* make `proxy run` disclose a local proxy token to a remote host.
@@ -35,1 +33,0 @@
-* The underlying proxy secret remains the HMAC key and is never sent upstream.
@@ -37,1 +34,0 @@
-export declare function createMetaProxyPolicyToken(proxyToken: string, policy: WorktreePolicy, worktree?: string, now?: number): string;
diff --git a/dist/meta-proxy.js b/dist/meta-proxy.js
--- a/dist/meta-proxy.js
+++ b/dist/meta-proxy.js
@@ -77,1 +77,0 @@
-export function metaProxyTokenPath(home = homedir()) {
@@ -83,1 +82,0 @@
-* make `proxy run` disclose a local proxy token to a remote host.
@@ -105,1 +103,0 @@
-token = readFileSync(metaProxyTokenPath(home), 'utf8').trim();
@@ -109,1 +106,0 @@
-throw new Error('Meta-Proxy token is unavailable. Start Meta-Proxy once to create its local token.');
@@ -123,1 +119,0 @@
-* The underlying proxy secret remains the HMAC key and is never sent upstream.
diff --git a/dist/secrets.js b/dist/secrets.js
--- a/dist/secrets.js
+++ b/dist/secrets.js
@@ -112,1 +112,0 @@
-lines.push(`  PASS secret '${secretName}' exists`);

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.4.21. 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: metaharness: Credential pass-through configuration (KillerSkills verdict caution 64)

## 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 64/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/npm/metaharness/0.4.21

## Findings with a location
- `dist/meta-proxy.d.ts:21` · Credential pass-through configuration — dist/meta-proxy.d.ts:21 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.d.ts:25` · Credential pass-through configuration — dist/meta-proxy.d.ts:25 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.js:105` · Credential pass-through configuration — dist/meta-proxy.js:105 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.d.ts:37` · Credential pass-through configuration — dist/meta-proxy.d.ts:37 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/secrets.js:112` · Credential pass-through configuration — dist/secrets.js:112 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.js:77` · Credential pass-through configuration — dist/meta-proxy.js:77 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.js:83` · Credential pass-through configuration — dist/meta-proxy.js:83 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.js:123` · Credential pass-through configuration — dist/meta-proxy.js:123 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.js:109` · Credential pass-through configuration — dist/meta-proxy.js:109 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/meta-proxy.d.ts:35` · Credential pass-through configuration — dist/meta-proxy.d.ts:35 · Credential pass-through configuration — agent relays sensitive credentials

## Proposed change
The attached patch removes the 10 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._