Salem AI / Developer tools / v0.1.0
AI Agent
Security Scanner.
Inspect your MCP configuration before giving an agent access.
Local checks for Cursor and VS Code. Review launch commands, declared permissions, transport and credential handling.
01 / Configuration
Start with a sanitized file.
Remove API keys, passwords, tokens and private customer data first. The scanner does not need them. The selected file is processed locally and is never submitted with a consultation request.
Understand the result
Configuration evidence.
Clear boundaries.
The scanner does not execute commands or contact your MCP servers. It cannot verify runtime authorization, successful prompt injection or actual data leakage.
Read the methodology ↗MCP security review guide ↗For teams deploying agents
Review the system.
Not just the file.
A configuration check cannot inspect how your agent handles real tool calls, customer data or approvals. Tell us what you are deploying to discuss a scoped security assessment.
- Agent and tool permission boundaries
- Data access and outbound transfers
- Approval controls and deployment risks
- Remediation priorities and retesting scope
Scope, price and delivery are agreed before any paid work. No automated subscription.
Discuss in WhatsApp ↗