Security & DLP proxy for MCP: tool-poisoning scans, PII redaction on tool args/results. Beta.
claude mcp add mcp-gateway --transport http https://api.aisecuritygateway.ai/mcp
~/.cursor/mcp.json
{
"mcpServers": {
"mcp-gateway": {
"url": "https://api.aisecuritygateway.ai/mcp"
}
}
}
~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"mcp-gateway": {
"serverUrl": "https://api.aisecuritygateway.ai/mcp"
}
}
}
.vscode/mcp.json
{
"servers": {
"mcp-gateway": {
"type": "http",
"url": "https://api.aisecuritygateway.ai/mcp"
}
}
}
cline_mcp_settings.json
{
"mcpServers": {
"mcp-gateway": {
"url": "https://api.aisecuritygateway.ai/mcp"
}
}
}
claude_desktop_config.json
{
"mcpServers": {
"mcp-gateway": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://api.aisecuritygateway.ai/mcp"
]
}
}
}
| Transport | URL | Auth headers |
|---|---|---|
| streamable-http | https://api.aisecuritygateway.ai/mcp |
Authorization: •••••• (required)
|
Reliable across a long session; no dropped connections.
Fast responses, sensible error messages when inputs were off.
Great docs mapping to real tool behavior. Auth was painless.
Fast responses, sensible error messages when inputs were off.
Did the job but rate limits kicked in under heavy use.
Great docs mapping to real tool behavior. Auth was painless.
POST /mcp