Propose a patch

Evidence

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

diff --git a/reference/creative-layouts.md b/reference/creative-layouts.md
--- a/reference/creative-layouts.md
+++ b/reference/creative-layouts.md
@@ -72,1 +72,0 @@
-叠加标题(run 级字号 + 颜色差异)
diff --git a/scripts/test/fixtures/slides/advice-for-newcomers.html b/scripts/test/fixtures/slides/advice-for-newcomers.html
--- a/scripts/test/fixtures/slides/advice-for-newcomers.html
+++ b/scripts/test/fixtures/slides/advice-for-newcomers.html
@@ -6,1 +6,0 @@
-<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800;900&family=Noto+Sans+SC:wght@300;400;500;700;900&display=swap" rel="stylesheet">
diff --git a/scripts/test/fixtures/slides/ai-trends-2026.html b/scripts/test/fixtures/slides/ai-trends-2026.html
--- a/scripts/test/fixtures/slides/ai-trends-2026.html
+++ b/scripts/test/fixtures/slides/ai-trends-2026.html
@@ -6,1 +6,0 @@
-<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800;900&family=Noto+Sans+SC:wght@300;400;500;700;900&display=swap" rel="stylesheet">
diff --git a/scripts/test/fixtures/slides/93-p2-sup-sub-geo.html b/scripts/test/fixtures/slides/93-p2-sup-sub-geo.html
--- a/scripts/test/fixtures/slides/93-p2-sup-sub-geo.html
+++ b/scripts/test/fixtures/slides/93-p2-sup-sub-geo.html
@@ -2,1 +2,0 @@
-================================================================ P2 特性夹具页(2026-07-27) 覆盖:上下标(<sup>/<sub>/vertical-align)、run 级字距、 预设几何(data-shape:triangle/right
diff --git a/scripts/test/fixtures/slides/90-p0-features.html b/scripts/test/fixtures/slides/90-p0-features.html
--- a/scripts/test/fixtures/slides/90-p0-features.html
+++ b/scripts/test/fixtures/slides/90-p0-features.html
@@ -2,1 +2,0 @@
-================================================================ P0 特性夹具页(2026-07-27) 覆盖:斜体/下划线/删除线/超链接、run 级字体、text-transform、 虚线/点线边框(统一+逐边)、真实 box-shadow、纯旋转
@@ -35,1 +34,0 @@
-2. run 级字体:中文基座混排西文 run
diff --git a/assets/snippets/kpi-card.html b/assets/snippets/kpi-card.html
--- a/assets/snippets/kpi-card.html
+++ b/assets/snippets/kpi-card.html
@@ -1,1 +1,0 @@
-kpi-card.html — KPI 卡片(渐变背景 + 大数字 + 上下标 + run 级字距) 用法:复制到页面,替换 {{参数}}。⚠️ 仅仪表盘原型(阅读/混合档);演讲档 KPI 用 stat-number 裸排(原型 17)。 需要 nativeGradient:true 才走原生渐变(否则截图)。 参数
diff --git a/scripts/test/fixtures/slides/four-revolutions.html b/scripts/test/fixtures/slides/four-revolutions.html
--- a/scripts/test/fixtures/slides/four-revolutions.html
+++ b/scripts/test/fixtures/slides/four-revolutions.html
@@ -6,1 +6,0 @@
-<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800;900&family=Noto+Sans+SC:wght@300;400;500;700;900&display=swap" rel="stylesheet">

Zero-context hunks: apply with git apply --unidiff-zero fix.patch on the tagged source of 2.1.0. 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: html-slides-to-pptx: HTML comment carrying imperative instructions to the agent (KillerSkills verdict caution 58)

## 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 58/100**. Summary: static analysis only — behavior not yet observed.

- See the findings below.

Full trace and findings: https://killerskills.ai/skills/clawhub/html-slides-to-pptx/2.1.0

## Findings with a location
- `reference/creative-layouts.md:72` · HTML comment carrying imperative instructions to the agent — reference/creative-layouts.md:72 · HTML comment carrying imperative instructions to the agent
- `scripts/test/fixtures/slides/advice-for-newcomers.html:6` · URL with embedded credentials — scripts/test/fixtures/slides/advice-for-newcomers.html:6 · URL with embedded credentials — may be phishing or credential leaking
- `scripts/test/fixtures/slides/ai-trends-2026.html:6` · URL with embedded credentials — scripts/test/fixtures/slides/ai-trends-2026.html:6 · URL with embedded credentials — may be phishing or credential leaking
- `scripts/test/fixtures/slides/93-p2-sup-sub-geo.html:2` · HTML comment carrying imperative instructions to the agent — scripts/test/fixtures/slides/93-p2-sup-sub-geo.html:2 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file
- `scripts/test/fixtures/slides/90-p0-features.html:35` · HTML comment carrying imperative instructions to the agent — scripts/test/fixtures/slides/90-p0-features.html:35 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file
- `assets/snippets/kpi-card.html:1` · HTML comment carrying imperative instructions to the agent — assets/snippets/kpi-card.html:1 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file
- `scripts/test/fixtures/slides/four-revolutions.html:6` · URL with embedded credentials — scripts/test/fixtures/slides/four-revolutions.html:6 · URL with embedded credentials — may be phishing or credential leaking
- `scripts/test/fixtures/slides/90-p0-features.html:2` · HTML comment carrying imperative instructions to the agent — scripts/test/fixtures/slides/90-p0-features.html:2 · HTML comment carrying imperative instructions to the agent · informational: not an instruction file

## Proposed change
The attached patch removes the 8 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._