Your config decides what
runs on your machine.
MCP config files control which servers launch, what commands they run, and what secrets they access. A single misconfigured entry can pipe curl to shell, expose API keys, or open your system to the network. Scandar audits your config before anything starts.
WHAT IT DETECTS
6 threat categories for config files.
Insecure Transport
Detects HTTP connections to remote servers, missing TLS, and servers binding to 0.0.0.0 (exposed to the network).
Dangerous Commands
Flags shell interpreters (bash -c), sudo, curl piped to shell, netcat, and other unsafe command patterns.
Untrusted Sources
Catches raw GitHub URLs, git+ repos, raw IP addresses, non-standard ports, and npx -y auto-install patterns.
Secret Exposure
Identifies hardcoded API keys, bearer tokens, AWS credentials, and secrets that should use env var interpolation.
Dangerous Combinations
Cross-server analysis: file access + network = exfiltration risk, credentials + network = theft risk, shell + others.
Misconfiguration
Detects servers with no command or URL, empty args on runners like npx/uvx, and invalid JSON configurations.
SUPPORTED FORMATS
Works with every major MCP client.
Claude Desktopclaude_desktop_config.json
Claude Code.mcp.json
Cursormcp.json
Windsurfmcp_config.json
VS Codemcp.json
CustomAny JSON with mcpServers key
HOW IT WORKS
Layer 1 pattern analysis. Instant results.
Config files are structured JSON — no LLM needed. Scandar parses your config, inspects every server entry for dangerous patterns, checks secrets against env var interpolation, and analyzes cross-server capabilities for dangerous combinations. Results in milliseconds.
1
Parse
Detects flat and nested mcpServers formats
2
Inspect
25+ rules across 6 threat categories
3
Cross-check
Analyzes server capability combinations
4
Score
Trust score with per-finding severity
Scan Your MCP Config
Drop your config file or paste the JSON. Results in milliseconds.
Scan a Config FileENTERPRISE
Need fleet-wide AI security?
Scandar Overwatch gives you real-time visibility into every agent in your organization — policies, compliance reports, alert routing, and kill chain detection. Self-serve setup in 25 minutes.
Explore Overwatch →