Psychosynth

Psychosynth is an agent-native, on-chain psychometric data marketplace. It delivers high-variance, human-curated personality profiles and profile-conditioned behavioral responses to autonomous systems. Settle transactions query-by-query in USDC on Base using the x402 micro-payment protocol—completely free of human intermediaries.

System Architecture

graph TD
    A["Generator OS (DB Templates)"] --> B["LLM Synthesis (Flash/Sonnet)"]
    B --> C["Hooks Pipeline (validation & dedup)"]
    C --> D["Curation Queue (Review Desk)"]
    D --> E["Supabase DB Store"]
    E --> F["x402 Proxy Middleware (Base)"]
    F --> G["JSON API / MCP Server"]
    G --> H["Autonomous Agents (Eliza, OpenClaw, Hermes)"]

Related MCP server: happy-thoughts

Core Capabilities

1. v2 Psychometric Engine

Standard LLMs suffer from "average-model bias," producing homogeneous, risk-neutral responses in games and simulations. Psychosynth resolves this by delivering structured, high-variance datasets containing:

• Five-Factor Model (OCEAN): Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism.

• Dark Triad Traits: Machiavellianism, Narcissism, and Psychopathy.

• Prospect Theory Posture: Loss aversion coefficients ($\lambda$) and power utility exponents for gains ($\alpha$) and losses ($\beta$).

• Cognitive Reflection: System 1 (heuristic-driven) vs. System 2 (logical/deliberative) preferences, alongside Cognitive Reflection Test (CRT) scores.

2. Agent-Native Payments (x402)

Transactions settle query-by-query on the Base blockchain using the x402 standard:

• Gasless Settlement: Agents use their EVM wallets to sign gasless TransferWithAuthorization payloads (EIP-3009) in USDC.

• Proxy Middleware: The backend verifies the signature, broadcasts the settlement on-chain, and delivers the requested dataset in a single HTTP request loop.

3. Model Context Protocol (MCP) Server

Psychosynth exposes an MCP server to connect directly with autonomous agent runtimes.

• Supported Frameworks: ElizaOS, OpenClaw (via MCPorter), and Nous Research's Hermes Agent.

• Exposed Tools:

  • list_products: Discover available products, schemas, and pricing.

  • preview_records: Fetch free, deterministic samples to verify schema shape.

  • get_quote: Request an x402 payment quote without executing transactions.

  • query_records: Execute on-chain payments and retrieve full data payloads.

4. Interactive Lab OS Playground

Psychosynth hosts an ultra-premium, dark-luxury financial dashboard for human interaction and debugging at /playground:

• Real-Time Radar Chart: Dynamic SVG spider map showing 8 psychometric dimensions (OCEAN, Dark Triad, Prospect Theory) updating instantly as parameters change.

• Holographic Agent Passports: Generates beautiful, shareable glassmorphism identity cards with automated archetype classification and direct 1-click Twitter sharing.

• Multi-Agent Duel Arena: A live simulation pit where two agents with distinct profiles (e.g., Solana Degen vs. Risk-Averse Institution) negotiate a transaction, with System 1/2 reasoning streaming in real-time.

• x402 Protocol Inspector: Step-by-step interactive terminal visualizing the HTTP 402 payment challenge, EIP-712 signing, Base facilitator settlement, and payload decryption.

Setup & Installation

Prerequisites

• Node.js >= 18

• Supabase CLI (optional, for local DB development)

Quick Start

1. Clone the repository and install dependencies:

   git clone https://github.com/3esign/psychosynth.git
   cd psychosynth
   npm install

2. Configure environment variables:

   cp .env.example .env
   # Fill in Supabase keys, X402 payout address, and keys.

3. Deploy Supabase migrations and seed data:

   npx supabase db push
   # Seeds the base schema and version 2 generators

4. Launch the Next.js development server:

   npm run dev

MCP Server Configuration

Build the MCP server (npm run build inside mcp/), then register it with your agent by adding the following entry to your MCP client config (e.g. claude_desktop_config.json). Use the absolute path to the built server on your machine:

{
  "mcpServers": {
    "psychosynth": {
      "command": "node",
      "args": ["/absolute/path/to/psychosynth/mcp/dist/index.js"],
      "env": {
        "PSYCHOSYNTH_API_URL": "https://psychosynth.vercel.app",
        "BUYER_PRIVATE_KEY": "0xYourBuyerWalletPrivateKey",
        "BASE_RPC_URL": "https://mainnet.base.org"
      }
    }
  }
}

CLI & Scripts

• npm run typecheck — Runs TypeScript compiler checks without emitting files.

• npm run buyer-test — Simulates an end-to-end paid query against a running instance.

• npm run export:sft | export:dpo | export:reject-cls — Exports curation data into training formats.

Documentation

Full architectural specifications, master plans, and developer logs reside in the docs/ folder:

• MASTERPLAN.md — Product vision and economic structures.

• DISCOVERY.md — Framework-specific agent integration details.

• DEVELOPMENT.md — Detailed engineering specs and database schema designs.

Bankr Ecosystem Integration

Psychosynth speaks standard x402: agents sign a gasless USDC EIP-3009 TransferWithAuthorization on Base and the server settles it via an x402 facilitator (the facilitator broadcasts and pays gas). This is the payment shape Bankr platform wallets, x402-fetch, and most agent wallet layers produce automatically. Self-settled txHash payments (Base or Solana) remain supported as a fallback.

• Free agent preflight: GET /api/v1/discovery — products, live prices, tiers, payTo, and settlement methods in one call.

• Bankr skill: the submission package for the BankrBot/skills catalog lives in integrations/bankr-skills/psychosynth/ (SKILL.md + catalog.json + logo + references + scripts). Once merged it surfaces on skills.bankr.bot and installs with: install the psychosynth skill from https://github.com/BankrBot/skills/tree/main/psychosynth.

• Positioning: existing intelligence skills in that ecosystem analyze tokens; Psychosynth sells synthetic behavioral data — priors about how market participants act — for trading sims, counterparty modeling, and agent stress-testing.