Private SuperIntelligence

si.privatesuperintelligencev0.1.0Updated Oct 7, 2026

Private SuperIntelligence by Mitosis Labs: define the term, see privacy layers, join the waitlist.

Overview

AI-generated overview

Lets an assistant explain Private SuperIntelligence, browse its site content, and submit a waitlist registration on the user's behalf.

What it does
Exposes tools such as get_service_overview, get_privacy_layers, get_benchmarks, search_site, read_page and register_interest. The assistant can pull the product definition, privacy layers and benchmark material, search and read pages of the vendor's site, and register interest with contact details. A CLI and SDK offer the same operations, and registration supports a dry-run mode that validates without sending.
When to use it
Useful when someone wants an assistant to answer questions about this private AI concierge offering, or to look up its published material, without leaving the chat. The registration tool matters only if the user actually intends to join the waitlist.
Requirements
A remote Streamable HTTP endpoint; no API key is required. Clients that can only launch local servers need Node.js to run the npx stdio bridge. Network access to the vendor's site is needed for search and page reads.
Before you install
register_interest sends the user's email, job title and revenue band to the vendor, so it should only run after the person approves those exact details; the CLI prompts before sending and supports a dry-run. Search and page reads query the vendor's site.

Installation

In SourceWeft

  1. Open Private SuperIntelligence in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "private-superintelligence": {
      "type": "http",
      "url": "https://privatesuperintelligence.si/mcp"
    }
  }
}

README

Private SuperIntelligence for agents

Official SDK, CLI, MCP bridge, skills and plugin for Private SuperIntelligence, the private AI concierge from Mitosis Labs for enterprises, family offices and private clients.

Private superintelligence is AI with the reach of a superintelligent system that works for one owner, on that owner's information, inside boundaries the owner controls. Read the full definition.

Every surface is free and needs no API key.

Connect an agent

Remote MCP server (Streamable HTTP):

bash
claude mcp add --transport http private-superintelligence https://privatesuperintelligence.si/mcp

Clients that only launch local servers can use the stdio bridge:

json
{ "mcpServers": { "private-superintelligence": { "command": "npx", "args": ["-y", "private-superintelligence", "mcp"] } } }

Tools: get_service_overview, get_privacy_layers, get_benchmarks, search_site, read_page, register_interest.

CLI

bash
npx private-superintelligence define       # or install globally and run `psi define`psi aboutpsi layerspsi search "zero retention"psi register --email [email protected] --title "Chief Investment Officer" --revenue 250m-1b --dry-run

psi register shows the details and asks before it sends anything. --dry-run validates without sending.

SDK

bash
npm install private-superintelligence
ts
import { PrivateSuperIntelligence } from 'private-superintelligence';
const psi = new PrivateSuperIntelligence();const { text } = await psi.definition();const { results } = await psi.search('air-gapped deployment');
// Only after the person has approved these exact details:await psi.registerInterest(  { email: '[email protected]', jobTitle: 'Chief Investment Officer', annualRevenueBand: '250m-1b', userConfirmed: true },  { idempotencyKey: crypto.randomUUID() },);

Errors throw PrivateSuperIntelligenceError with the API's code, hint and field errors.

Skills and plugin

Install the skills with npx skills add OperatingSystem-1/private-superintelligence-agent.

Links

MIT licensed.

Source: README.md at commit b294704

Tools

0
Tool metadata has not been indexed yet.

Version history

1
  1. v0.1.0LatestOct 7, 2026