A ninety-second read of the body, not the language. Compressed into one word, one rhythm, one orientation. Any AI surface can load it with one call.
Every conversation with Claude, GPT, Gemini, or any agent starts cold. The model has no read on who it is talking to. It defaults to generic-helpful. Memory features paper over this by replaying chat history, which captures what the person said but never what the person is.
Noctara solves this with a behavioral identity primitive. The user takes a structured 90-second reading at noctaracorp.com. The architecture reads the rhythm of how they type, not what they type. The output is a compressed identity: one spine word, one rhythm, one force, one half-truth, one compliance lever. The user owns the read. Apps consume it with the user's consent via signed JWT or MCP server.
The reference implementation is at noctaracorp.com. The MCP server installs in Claude Desktop, Claude Code, Cursor, Cline, OpenClaw with one config block. The SDKs are @noctara/sdk (TypeScript / npm) and noctara (Python / PyPI). The MCP server is noctara-mcp-server (npm). The open specification is at noctaracorp.com/spec. The patent is US provisional application #64/048,624 (filed April 24, 2026, with the Rapacke Law Group).
import { Noctara } from "@noctara/sdk";
import Anthropic from "@anthropic-ai/sdk";
const noctara = new Noctara();
const claude = new Anthropic();
const { system_prompt } = await noctara.systemPrompt({
email: "user@example.com",
});
const msg = await claude.messages.create({
model: "claude-opus-4-7",
max_tokens: 1024,
system: system_prompt + "\n\n# Your task. You are this person's agent.",
messages: [{ role: "user", content: "What should I do tomorrow?" }],
});
The model now writes back in the user's rhythm. Not a default. Not a tone setting. The body's read of the person carrying into the language.
Every major AI lab is racing toward a personalization layer. They are racing through chat-history-as-memory, which is shallow. Noctara is the deeper primitive. It reads the body. It compresses. It hands the model a stamp of who the user is, not a transcript of what they said.
Personality is what people say about themselves. Identity is what the body says about them.
The behavioral identity primitive described in our specification is patent-protected. We are open to licensing the claims (1-7) to model providers who want to integrate at the system-prompt or memory layer. The architecture is designed for privacy-preserving integration: the user owns the read, opts in per app, and can revoke at any time from noctaracorp.com/identity.
For licensing inquiries: cole@noctaracorp.com.
Personal and non-commercial use is free. Commercial use up to 10,000 monthly active users is free under attribution. Above that, license terms are simple and posted at /developers. The SDKs are MIT. The patent does not encumber the SDK, only the primitive.