中文 EN

Scan Report

Scan New Files

✓ Trusted — Risk Score 5/100
Last scan:23 hr ago Rescan
5 /100
knowledge-card
Generate beautiful knowledge concept cards (知识概念卡片) from book summaries, notes, or topics as PNG images via Playwright screenshot
A straightforward knowledge-card PNG generator that uses Playwright to screenshot HTML templates. No malicious behavior, credential access, obfuscation, or undeclared capabilities detected.
Skill Nameknowledge-card
Duration35.6s
Enginepi
✓
Safe to install
No action needed. The skill is safe to use as described.

Findings 2 items

Severity Finding Location
Low
Playwright loaded from unversioned npx cache path Supply Chain
The script hardcodes the Playwright module path as /root/.npm/_npx/e41f203b7505f1fb/node_modules/playwright, bypassing any version declaration. No package.json or requirements.txt exists to pin the dependency version.
const { chromium } = require('/root/.npm/_npx/e41f203b7505f1fb/node_modules/playwright');
→ Add a package.json with a pinned Playwright version and a lock file to ensure reproducible builds.
 scripts/generate.js:11
Low
Network READ access not declared in SKILL.md Doc Mismatch
SKILL.md does not mention that the skill fetches Google Fonts from fonts.googleapis.com when generating cards. The network access is minor (font loading only) but not documented.
No declaration of network:READ in allowed-tools mapping
→ Add a note in SKILL.md that Google Fonts are fetched from CDN when rendering cards.
 SKILL.md:1
ResourceDeclaredInferredStatusEvidence
Filesystem WRITE WRITE ✓ Aligned scripts/generate.js:56 — fs.writeFileSync writes temp HTML; fs.writeFileSync wri…
Shell WRITE WRITE ✓ Aligned SKILL.md:60 — node scripts/generate.js (user-facing shell invocation); scripts/g…
Browser READ READ ✓ Aligned scripts/generate.js:71-74 — Playwright launches browser, navigates to file:// HT…
Network NONE READ ✓ Aligned scripts/generate.js:48 — @import url(https://fonts.googleapis.com/...) in genera…
Environment NONE NONE — No os.environ iteration or access
Clipboard NONE NONE — No clipboard access
Database NONE NONE — No database access
Skill Invoke NONE NONE — No recursive skill invocation

File Tree

2 files · 13.5 KB · 290 lines
JavaScript 1f · 149L Markdown 1f · 141L
├─ ▾ 📁 scripts
│ └─ 📜 generate.js JavaScript 149L · 9.0 KB
└─ 📝 SKILL.md Markdown 141L · 4.5 KB

Dependencies 2 items

PackageVersionSourceKnown VulnsNotes
playwright unversioned npx cache /root/.npm/_npx/e41f203b7505f1fb/node_modules/playwright No Loaded from hardcoded npx cache path with no version pinning in a manifest
google-fonts (fonts.googleapis.com) N/A CDN No Fetched at render time via CSS @import in generated HTML

Security Positives

✓ No shell command injection vectors — all arguments are CLI-parsed strings with no eval() or child_process execution
✓ No credential, token, or sensitive path access (no ~/.ssh, ~/.aws, .env, or os.environ iteration)
✓ No obfuscation techniques — no base64, atob, eval, or dynamic code construction
✓ No data exfiltration — output PNG is written locally and printed to stdout only
✓ No persistence mechanisms — no cron, startup scripts, or backdoor installation
✓ No recursive skill invocation or clipboard access
✓ HTML template is static with controlled CSS injection (theme colors are simple string interpolation, not script execution)
✓ Temporary files are created in os.tmpdir() and cleaned up after screenshot
✓ Doc-to-code behavior is consistent — SKILL.md accurately describes node CLI usage and Playwright screenshot workflow