Propose a patch

Evidence

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

diff --git a/.github/workflows/corpus-refresh.yml b/.github/workflows/corpus-refresh.yml
--- a/.github/workflows/corpus-refresh.yml
+++ b/.github/workflows/corpus-refresh.yml
@@ -108,1 +108,0 @@
-origin="https://x-access-token:${GH_TOKEN}@github.com/${REPO}.git"
diff --git a/mcp_for_ocp_graphql/auth.py b/mcp_for_ocp_graphql/auth.py
--- a/mcp_for_ocp_graphql/auth.py
+++ b/mcp_for_ocp_graphql/auth.py
@@ -132,1 +132,0 @@
-placeholder="Paste your token here" spellcheck="false" autocomplete="off">
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -63,1 +63,0 @@
-git remote set-url origin "https://x-access-token:${GH_TOKEN}@github.com/${GITHUB_REPOSITORY}.git"
diff --git a/.opencrane/chunks.json b/.opencrane/chunks.json
--- a/.opencrane/chunks.json
+++ b/.opencrane/chunks.json
@@ -1013,1 +1013,0 @@
-\n<div>\n [Image removed: image]\n <p style=\"text-align:center;margin-bottom: 1em;\"><em>Applications section</em></p>\n [Image removed: image]\n <p style=\"te
diff --git a/mcp_for_ocp_graphql/data/docs.json b/mcp_for_ocp_graphql/data/docs.json
--- a/mcp_for_ocp_graphql/data/docs.json
+++ b/mcp_for_ocp_graphql/data/docs.json
@@ -473,1 +473,0 @@
-\n<div>\n [Image removed: image]\n <p style=\"text-align:center;margin-bottom: 1em;\"><em>Applications section</em></p>\n [Image removed: image]\n <p style=\"te

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 0.12.0. 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: mcp-for-ocp-graphql: URL with embedded credentials (KillerSkills verdict caution 77)

## 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 77/100**. Summary: no network, touches nothing outside the project, writes no memory.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/pypi/mcp-for-ocp-graphql/0.12.0

## Findings with a location
- `.github/workflows/corpus-refresh.yml:108` · URL with embedded credentials — .github/workflows/corpus-refresh.yml:108 · URL with embedded credentials — may be phishing or credential leaking
- `mcp_for_ocp_graphql/auth.py:132` · Prompting user to enter sensitive credentials — mcp_for_ocp_graphql/auth.py:132 · Prompting user to enter sensitive credentials — social engineering
- `.github/workflows/release.yml:63` · URL with embedded credentials — .github/workflows/release.yml:63 · URL with embedded credentials — may be phishing or credential leaking
- `.opencrane/chunks.json:1013` · HTML comment carrying imperative instructions to the agent — .opencrane/chunks.json:1013 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file · informational: not an instruction file · informational: not an instr
- `mcp_for_ocp_graphql/data/docs.json:473` · HTML comment carrying imperative instructions to the agent — mcp_for_ocp_graphql/data/docs.json:473 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file · informational: not an instruction file · informational: n

## 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._