HiddenContent
Check a document for hidden text before your agent reads it. PDF, Office, RTF, HTML.
Install
Add this to your client's MCP configuration (Claude Desktop: claude_desktop_config.json; Cursor: .cursor/mcp.json; VS Code and others use the same "mcpServers" shape).
claude mcp add --transport http hiddencontent 'https://api.hiddencontent.ai/v1/mcp'
{
"mcpServers": {
"hiddencontent": {
"type": "http",
"url": "https://api.hiddencontent.ai/v1/mcp"
}
}
}
@hiddencontent/mcpclaude mcp add hiddencontent -e 'HCS_API_TOKEN=<your hcs_api_token>' -e 'HCS_URL=<HCS_URL>' -e 'HCS_CAPABILITIES=<HCS_CAPABILITIES>' -- npx -y @hiddencontent/mcp@0.4.15
{
"mcpServers": {
"hiddencontent": {
"command": "npx",
"args": [
"-y",
"@hiddencontent/mcp@0.4.15"
],
"env": {
"HCS_API_TOKEN": "<your hcs_api_token>",
"HCS_URL": "<HCS_URL>",
"HCS_CAPABILITIES": "<HCS_CAPABILITIES>"
}
}
}
}
Environment variables
HCS_API_TOKENrequired secret API token for the HiddenContent service. Get one at https://hiddencontent.ai/start?ref=mcp-registry — the server refuses to start without it and says so.HCS_URLBase URL of the service. Defaults to the hosted API; set it only when pointing at a self-hosted instance.HCS_CAPABILITIESComma-separated list of what your agent can actually do. The only accepted values are `tool-use`, `network-egress` and `code-execution`. These turn on the checks for concealed text that instructs a model to call a tool, send data or run code — declaring more can raise severity and never lowers it, and declaring nothing leaves those checks inert. Defaults to `tool-use` in the bundle install.
https://github.com/AeroSpark-ai/hiddencontent-mcp/releases/download/v0.4.15/hiddencontent.mcpbEnvironment variables
HCS_API_TOKENrequired secret API token for the HiddenContent service. Get one at https://hiddencontent.ai/start?ref=mcp-registry — the server refuses to start without it and says so.HCS_URLBase URL of the service. Defaults to the hosted API; set it only when pointing at a self-hosted instance.HCS_CAPABILITIESComma-separated list of what your agent can actually do. The only accepted values are `tool-use`, `network-egress` and `code-execution`. These turn on the checks for concealed text that instructs a model to call a tool, send data or run code — declaring more can raise severity and never lowers it, and declaring nothing leaves those checks inert. Defaults to `tool-use` in the bundle install.
Frequently asked questions
What is HiddenContent?
Check a document for hidden text before your agent reads it. PDF, Office, RTF, HTML.
How do I install HiddenContent?
It runs as a remote server. In Claude Code: claude mcp add --transport http hiddencontent 'https://api.hiddencontent.ai/v1/mcp'. For other clients, paste the JSON configuration from this page into their MCP settings.
Does HiddenContent need an API key?
Yes. It expects HCS_API_TOKEN, which you set in its configuration.
Is HiddenContent open source?
Its code is public on GitHub, but the repository does not declare a standard license.
Who maintains HiddenContent?
It is published by AeroSpark-ai. AI.tn lists it but does not maintain or endorse it.
From the official MCP Registry. Listing is not an endorsement: review a server's code and permissions before giving it access to your data.