Verified-answer AI assistant: answers only from 33 owner-approved facts, never invents.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Ask Cynthia a question. She answers ONLY from her verified knowledge, returning the verified answer verbatim labelled "Verified answer". If the question is not in her verified knowledge, she honestly abstains instead of guessing. Args: question: The question to ask Cynthia, in plain English.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http cynthia https://cynthia-mcp-server.onrender.com/mcp
Ask Cynthia a question. She answers ONLY from her verified knowledge, returning the verified answer verbatim labelled "Verified answer". If the question is not in her verified knowledge, she honestly abstains instead of guessing. Args: question: The question to ask Cynthia, in plain English.
| Argument | Type | Description |
|---|---|---|
| question* | string |
{
"type": "object",
"title": "ask_cynthiaArguments",
"required": [
"question"
],
"properties": {
"question": {
"type": "string",
"title": "Question"
}
}
}Describe what Cynthia can do and the rules she follows (verified answers only, owner approval on risky actions, free-only building, subscription guard, and more).
No arguments.
{
"type": "object",
"title": "cynthia_capabilitiesArguments",
"properties": {}
}Report the Cynthia MCP server version, the number of verified facts in her knowledge store, and a health check.
No arguments.
{
"type": "object",
"title": "cynthia_statusArguments",
"properties": {}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 18, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.