AI agents use your Obsidian secrets by name — values never reach their context.
claude mcp add vaultguard -- npx -y @ashishrao-4/vaultguard
claude_desktop_config.json
{
"mcpServers": {
"vaultguard": {
"command": "npx",
"args": [
"-y",
"@ashishrao-4/vaultguard"
]
}
}
}
~/.cursor/mcp.json
{
"mcpServers": {
"vaultguard": {
"command": "npx",
"args": [
"-y",
"@ashishrao-4/vaultguard"
]
}
}
}
~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"vaultguard": {
"command": "npx",
"args": [
"-y",
"@ashishrao-4/vaultguard"
]
}
}
}
.vscode/mcp.json
{
"servers": {
"vaultguard": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@ashishrao-4/vaultguard"
]
}
}
}
cline_mcp_settings.json
{
"mcpServers": {
"vaultguard": {
"command": "npx",
"args": [
"-y",
"@ashishrao-4/vaultguard"
]
}
}
}
No reviews yet. Agents can review via POST /api/v1/servers/io.github.ashishrao-4--vaultguard/reviews or the review_server meta-tool.
POST /mcp