@atlisp/mcp@1.8.25
MCP Server for @lisp on CAD,support AutoCAD/GstarCAD/ZWCAD/BricsCAD or CAD platform compatible with AutoLISP
AI Security Review
scanned 3mo ago · by lpm-firewall-aiNo concrete malicious install-time behavior, credential theft, or stealth persistence was found. The runtime plugin lifecycle and broad network/CAD capability warrant a warning rather than a block.
Decision evidence
public snapshot- `dist/atlisp-mcp.js` starts an HTTP/WebSocket MCP service on configurable host, default `0.0.0.0`.
- `dist/atlisp-mcp.js` auto-loads and watches executable plugins from `~/.atlisp/mcp-plugins` at runtime.
- `dist/atlisp-mcp.js` posts periodic hostname/CAD/system health metadata to `https://atlisp.cn/api/mcp/health`.
- `dist/handlers/skill-handlers.js` explicitly downloads selected registry skills into `~/.atlisp/skills`.
- `dist/cad-worker.js` loads the bundled `dist/cad-rothelper.dll` through `edge-js` for CAD integration.
- `package.json` has no `preinstall`, `install`, or `postinstall` lifecycle hook; only `prepublishOnly` builds publisher artifacts.
- The child process only launches the bundled `cad-worker.js` for CAD IPC.
- No source evidence of credential harvesting, reading SSH/npm/AI-agent configuration, or secret exfiltration.
- Plugin and skill writes are confined to package-owned `~/.atlisp` paths and require runtime or explicit tool use.
Source & flagged code
4 flagged · loading sourcePackage source references child process execution.
dist/atlisp-mcp.jsView on unpkg · L543Source launches a detached bundled service that exposes a broad-bound HTTP listener.
dist/atlisp-mcp.jsView on unpkg · L77Findings
Affected versions and remediation
This report applies to @atlisp/mcp@1.8.25.
See all versions of @atlisp/mcp for other recorded verdicts.
- Review the evidence and your use of @atlisp/mcp@1.8.25 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.