Pangea MCP proxy

Protect communications between a client and any MCP server. Now with 99% less
prompt injection! The Pangea MCP proxy allows any MCP client to secure the
messages it sends and receives to/from an MCP server, using the
[Pangea AI Guard][] service to guard tools' inputs and outputs.

What it does: protect users from common threat vectors by running all MCP I/O
through Pangea AI Guard, which blocks:

- Prompt injections (yes, even the ones wrapped in a riddle)
- Malicious links, IPs, domains (via CrowdStrike, DomainTools, WhoisXML threat intel)
- 50 types of confidential information and PII
- 10 content filters, including toxicity, self harm, violence, and filtering by topic
- Support for 104 spoken languages

Bonus: It stores your AI Guard token safely in Pangea Vault, with automatic
rotation.

Extra bonus: Each request to AI Guard and its detection results are logged to
your Secure Audit Log, giving you an immutable trail of activity for audits,
debugging, and incident response.

Prerequisites

- Node.js v22.15.0 or greater.
- A Pangea API token with access to AI Guard. This token needs to be stored in
Pangea Vault. See [Service Tokens][] for documentation on how to create and
manage Pangea API tokens.
- A Pangea API token with access to Vault. This will be used to fetch the above
token at runtime.

Usage

In an existing stdio-based MCP server configuration like the following:

``json
{
"mcpServers": {
"qrcode": {
"command": "npx",
"args": ["-y", "@jwalsh/mcp-server-qrcode"]
}
}
}
`

Wrap the original command with npx -y @pangeacyber/mcp-proxy and add
environment variables:

`json
{
"mcpServers": {
"qrcode": {
"command": "npx",
"args": [
"-y",
"@pangeacyber/mcp-proxy",
"--",
"npx",
"-y",
"@jwalsh/mcp-server-qrcode"
],
"env": {
"PANGEA_VAULT_TOKEN": "pts_00000000000000000000000000000000",
"PANGEA_VAULT_ITEM_ID": "pvi_00000000000000000000000000000000"
}
}
}
}
`

1. Update the PANGEA_VAULT_TOKEN value to the Pangea Vault API token.
1. Update the PANGEA_VAULT_ITEM_ID value to the Vault item ID that contains
the Pangea AI Guard API token.

For remote servers using HTTP or SSE, use [mcp-remote][] to turn them into stdio
servers:

`json
{
"mcpServers": {
"proxied": {
"command": "npx",
"args": [
"-y",
"@pangeacyber/mcp-proxy",
"--",
"npx",
"-y",
"mcp-remote",
"https://remote.mcp.server/sse"
],
"env": {
"PANGEA_VAULT_TOKEN": "pts_00000000000000000000000000000000",
"PANGEA_VAULT_ITEM_ID": "pvi_00000000000000000000000000000000"
}
}
}
}
`

$3

To identify the calling app by ID in Pangea, set the APP_ID environment
variable.

$3

To identify the calling app by name in Pangea, set the APP_NAME environment
variable.

$3

To use a Pangea base URL other than the default
https://{SERVICE_NAME}.aws.us.pangea.cloud, set the PANGEA_BASE_URL_TEMPLATE
environment variable to a custom template (e.g. https://{SERVICE_NAME}.dev.pangea.cloud`).

[Pangea AI Guard]: https://pangea.cloud/docs/ai-guard/
[Service Tokens]: https://pangea.cloud/docs/admin-guide/projects/credentials#service-tokens
[mcp-remote]: https://github.com/geelen/mcp-remote