Propose a patch

Evidence

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

diff --git a/dist/agents/command.mjs b/dist/agents/command.mjs
--- a/dist/agents/command.mjs
+++ b/dist/agents/command.mjs
@@ -16,1 +16,0 @@
-import { proxyToken, startProxy } from "./proxy.mjs";
@@ -164,1 +163,0 @@
-const token = proxyToken();
diff --git a/README.md b/README.md
--- a/README.md
+++ b/README.md
@@ -189,1 +189,0 @@
-<summary><strong>Development and contributing</strong></summary> The CLI lives in `cli/` in the Phantom repo, which uses pnpm. Install once at the repo root, th
diff --git a/dist/agents/proxy.mjs b/dist/agents/proxy.mjs
--- a/dist/agents/proxy.mjs
+++ b/dist/agents/proxy.mjs
@@ -12,1 +12,0 @@
-* it started: each request must carry the run's token ({@link proxyToken}),
@@ -18,1 +17,0 @@
-export function proxyToken() {
@@ -46,1 +44,0 @@
-* @param opts.token The run's secret ({@link proxyToken}). A request without it, as `Authorization: Bearer` or `x-api-key`, gets 401.

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.6.0. 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: cli: Credential pass-through configuration (KillerSkills verdict caution 67)

## What the honeypot observed
KillerSkills ran **[email protected]** in an isolated sandbox (decoy credentials, recording proxy, syscall trace; engine 1.8.1) and scored it **CAUTION 67/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/npm/cli/1.6.0

## Findings with a location
- `dist/agents/command.mjs:16` · Credential pass-through configuration — dist/agents/command.mjs:16 · Credential pass-through configuration — agent relays sensitive credentials
- `README.md:189` · Collapsed <details> block carrying instructions — README.md:189 · Collapsed <details> block carrying instructions
- `dist/agents/proxy.mjs:18` · Credential pass-through configuration — dist/agents/proxy.mjs:18 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/agents/command.mjs:164` · Credential pass-through configuration — dist/agents/command.mjs:164 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/agents/proxy.mjs:46` · Credential pass-through configuration — dist/agents/proxy.mjs:46 · Credential pass-through configuration — agent relays sensitive credentials
- `dist/agents/proxy.mjs:12` · Credential pass-through configuration — dist/agents/proxy.mjs:12 · Credential pass-through configuration — agent relays sensitive credentials

## Proposed change
The attached patch removes the 6 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._