registry / @memnox/mcp / 2.0.11

@memnox/mcp@2.0.11

Memnox MCP server — organizational memory and context intelligence for AI coding tools and CI/CD pipelines

AI Security Review

scanned 22h ago · by lpm-firewall-ai

A crafted watched filename can escape the quoted git command and execute shell code. The MCP watcher may start automatically after a project-scoped tool call; standalone watch is explicit.

Static reason

No blocking static signals were detected.

Trigger

Run memnox watch, or invoke a project-selection MCP tool with project_id and cause a crafted filename to change.

Impact

Potential arbitrary command execution as the MCP/CLI user; STOP verdicts can also revert local changes.

Mechanism

unsanitized filename interpolation into execSync shell commands

Rationale

No install-time malware or covert exfiltration was found, but the watcher exposes a concrete shell-injection RCE path and automatic destructive revert behavior. Treat as a critical vulnerability requiring a warning, not a malware block.

Evidence

package.jsondist/index.jsdist/local-watcher.jsdist/commands/watch.jsdist/client.jsdist/commands/setup.js

Network endpoints1

api.memnox.com

Decision evidence

public snapshot

AI called this Suspicious at 92.0% confidence as Critical Vulnerability with false-positive risk.

@memnox/mcp@2.0.11

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code