@roarkanalytics/sdk-mcp@3.18.0
The official MCP Server for the Roark API
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThe package presents a real, externally influenced agent-execution surface, so it warrants a warning. Its behavior is explicit and constrained, and source inspection found no concrete malware chain.
Decision evidence
public snapshot- The MCP server fetches runtime instructions from an external Stainless endpoint when it starts.
- The enabled execute tool evaluates caller-supplied TypeScript and invokes it with an authenticated Roark client.
- Local execution grants the worker network access to the configured Roark API host and environment access.
- package.json has no preinstall, install, postinstall, or self-dependency.
- The normal CLI defaults code execution to local and rejects the retired remote sandbox mode.
- The local worker restricts network access to the configured API hostname; no covert file harvesting, persistence, or destructive action was found.
Source & flagged code
12 flagged · loading sourceSource file is highly similar to a previously finalized malicious package; route for source-aware review.
instructions.jsView on unpkgPackage source references dynamic require/import behavior.
instructions.jsView on unpkg · L7Findings
Affected versions and remediation
This report applies to @roarkanalytics/sdk-mcp@3.18.0.
See all versions of @roarkanalytics/sdk-mcp for other recorded verdicts.
- Review the evidence and your use of @roarkanalytics/sdk-mcp@3.18.0 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.