中文 EN

Scan Report

Scan New Files

✓ Trusted — Risk Score 5/100
Last scan:1 day ago Rescan
5 /100
wechat-cover
Generate WeChat official account cover images with proper 2.35:1 aspect ratio. Supports OpenAI DALL-E and Gemini image generation with customizable styles.
This is a legitimate WeChat cover image generator with no malicious behavior found. All capabilities are declared, dependencies are version-pinned, and no credential theft, data exfiltration, obfuscation, or unauthorized access was detected.
Skill Namewechat-cover
Duration45.7s
Enginepi
✓
Safe to install
This skill is safe to use. No action required.

Findings 2 items

Severity Finding Location
Info
Incomplete capability declaration in SKILL.md Doc Mismatch
SKILL.md does not explicitly declare filesystem:READ (for settings.json) or filesystem:WRITE (for image output), though both are clearly implied by usage examples and the output section. The environment variable access is properly documented.
SKILL.md describes image output in the Output section but omits explicit filesystem capability declarations.
→ Add 'filesystem: READ (for settings.json)' and 'filesystem: WRITE (for generated images)' to the skill declaration header for full transparency.
 SKILL.md:1
Info
Dependencies declared in inline script header Supply Chain
generate.py declares dependencies with minimum version pins ('openai>=1.12.0', 'pillow>=10.0.0', 'google-genai>=0.8.0') in its inline script header. While these are lower-bound pins rather than exact versions, they are declared and auditable.
requires-python = ">=3.11"
dependencies = ["openai>=1.12.0", "pillow>=10.0.0", "google-genai>=0.8.0"]
→ Consider using exact pinned versions (e.g., openai==1.12.0) for reproducible builds.
 scripts/generate.py:2
ResourceDeclaredInferredStatusEvidence
Filesystem NONE READ ✓ Aligned SKILL.md does not mention reading settings.json or its role. However, file reads…
Filesystem NONE WRITE ✓ Aligned SKILL.md usage examples show --output-dir and --filename parameters but does not…
Network NONE READ ✓ Aligned SKILL.md declares OpenAI and Gemini API calls in the Configuration and Requireme…
Environment NONE READ ✓ Aligned SKILL.md declares OPENAI_API_KEY, OPENAI_BASE_URL, and GEMINI_API_KEY environmen…
Shell NONE NONE — No shell execution found. 'uv run' is a Python script runner, not shell executio…
3 findings
🔗
Medium External URL 外部 URL
https://your-proxy.com/v1
SKILL.md:37
🔗
Medium External URL 外部 URL
https://platform.openai.com/api-keys
SKILL.md:152
🔗
Medium External URL 外部 URL
https://aistudio.google.com/app/apikey
SKILL.md:153

File Tree

6 files · 33.0 KB · 985 lines
Python 4f · 818L Markdown 1f · 153L JSON 1f · 14L
├─ ▾ 📁 scripts
│ ├─ 🐍 gemini_client.py Python 99L · 3.1 KB
│ ├─ 🐍 generate.py Python 472L · 16.5 KB
│ ├─ 🐍 image_client.py Python 83L · 2.1 KB
│ └─ 🐍 openai_client.py Python 164L · 5.3 KB
├─ 📋 settings.json JSON 14L · 314 B
└─ 📝 SKILL.md Markdown 153L · 5.7 KB

Dependencies 3 items

PackageVersionSourceKnown VulnsNotes
openai >=1.12.0 inline script header No Lower-bound pin; should consider exact version pin
pillow >=10.0.0 inline script header No Lower-bound pin; well-established image processing library
google-genai >=0.8.0 inline script header No Lower-bound pin; Google's official Gemini SDK

Security Positives

✓ All dependencies are declared with version constraints in the inline script header
✓ API keys are only used to authenticate with the intended provider — no credential exfiltration
✓ No shell execution, subprocess, os.system, or any form of command injection
✓ No access to sensitive paths (~/.ssh, ~/.aws, .env, etc.)
✓ No obfuscation (no base64 decode + execute patterns, no eval, no dynamic code generation)
✓ No supply chain threats — all packages are from well-known, established sources (openai, pillow, google-genai)
✓ No C2 communication or data exfiltration — network calls are exclusively to documented AI provider APIs
✓ Image output is user-controlled and limited to the specified output directory
✓ Settings.json contains only placeholder API keys (not real credentials)
✓ Clean code with no suspicious patterns, hidden functionality, or shadow features
✓ Error handling is appropriate and does not leak sensitive information