Propose a patch

Evidence

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

diff --git a/dist/infraone.js b/dist/infraone.js
--- a/dist/infraone.js
+++ b/dist/infraone.js
@@ -335,1 +335,0 @@
-…r and store the credential").option("--no-browser", "print the sign-in link and paste the token instead").option("--json", "print the result as JSON").helpGroup(general).actio…
diff --git a/dist/chunk-CH3WOVOV.js b/dist/chunk-CH3WOVOV.js
--- a/dist/chunk-CH3WOVOV.js
+++ b/dist/chunk-CH3WOVOV.js
@@ -2219,1 +2219,0 @@
-`Open ${baseUrl}/cli/login in a browser, authorize, and paste the token below (an outside expert pastes the pxc_ credential minted in their expert view instead).`
diff --git a/dist/chunk-RZL23EQ6.js b/dist/chunk-RZL23EQ6.js
--- a/dist/chunk-RZL23EQ6.js
+++ b/dist/chunk-RZL23EQ6.js
@@ -228,1 +228,0 @@
-return { ok: false, message: "--no-browser needs a terminal to paste the token into; for scripts, set INFRAONE_TOKEN" };
@@ -230,1 +229,0 @@
-notify(`Open ${loginPageUrl(config)} in a browser, authorize, and paste the token below.`);
diff --git a/README.md b/README.md
--- a/README.md
+++ b/README.md
@@ -1134,1 +1134,0 @@
-- `harness steer <job-id> <instruction>` posts the instruction in the job's

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 1.32.0. 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: harness-cli: Prompting user to enter sensitive credentials (KillerSkills verdict caution 68)

## 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 68/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/npm/harness-cli/1.32.0

## Findings with a location
- `dist/infraone.js:335` · Prompting user to enter sensitive credentials — dist/infraone.js:335 · Prompting user to enter sensitive credentials — social engineering
- `dist/chunk-CH3WOVOV.js:2219` · Prompting user to enter sensitive credentials — dist/chunk-CH3WOVOV.js:2219 · Prompting user to enter sensitive credentials — social engineering
- `dist/chunk-RZL23EQ6.js:228` · Prompting user to enter sensitive credentials — dist/chunk-RZL23EQ6.js:228 · Prompting user to enter sensitive credentials — social engineering
- `dist/chunk-RZL23EQ6.js:230` · Prompting user to enter sensitive credentials — dist/chunk-RZL23EQ6.js:230 · Prompting user to enter sensitive credentials — social engineering
- `README.md:1134` · Fake instruction boundary tag — README.md:1134 · Fake instruction boundary tag · informational: not an instruction file

## Proposed change
The attached patch removes the 5 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._