Trusted — Risk Score 5/100
Last scan:2 days ago Rescan
5 /100
zhipu-coding-plan-mcp
智谱 AI 视觉、搜索与生图工具集
This is a legitimate Zhipu AI (智谱) MCP tool integration that provides image analysis, web search, and AI generation capabilities. All credential access and execution is declared and necessary for the documented functionality.
Skill Namezhipu-coding-plan-mcp
Duration27.5s
Enginepi
Safe to install
No action needed. The skill follows security best practices by reading API keys from a config file rather than hardcoding them.
ResourceDeclaredInferredStatusEvidence
Shell WRITE WRITE ✓ Aligned execFileSync('npx', ...) in scripts/zai-mcp.js:42 - declared in SKILL.md metadat…
Filesystem READ READ ✓ Aligned readFileSync for auth-profiles.json and mcporter.json
Network READ READ ✓ Aligned HTTPS calls to open.bigmodel.cn for AI services
7 findings
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn
SKILL.md:47
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/paas/v4/images/generations
SKILL.md:165
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/paas/v4/videos/generations
SKILL.md:184
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/paas/v4/async-result/$TASK_ID
SKILL.md:211
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/mcp/web_search_prime/mcp
mcporter.json:14
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/mcp/web_reader/mcp
mcporter.json:19
🔗
Medium External URL 外部 URL
https://open.bigmodel.cn/api/mcp/zread/mcp
mcporter.json:24

File Tree

3 files · 11.6 KB · 348 lines
Markdown 1f · 271L JavaScript 1f · 49L JSON 1f · 28L
├─ 📁 scripts
│ └─ 📜 zai-mcp.js JavaScript 49L · 1.5 KB
├─ 📋 mcporter.json JSON 28L · 712 B
└─ 📝 SKILL.md Markdown 271L · 9.4 KB

Dependencies 2 items

PackageVersionSourceKnown VulnsNotes
npx * system No Required dependency declared in metadata
mcporter * npm No MCP server orchestrator

Security Positives

✓ API key is read dynamically from config file, not hardcoded
✓ npx dependency declared in SKILL.md metadata
✓ All network calls go to official Zhipu API endpoint (open.bigmodel.cn)
✓ No credential exfiltration - keys stay local
✓ Script is straightforward with no obfuscation or base64 encoding
✓ MCP servers use proper HTTP bearer token authentication