@noetic-tools/core

Core primitives for the Noetic agent framework: step types, the interpreter, the runtime, the memory layer contract, and built-in patterns.

Install

npm install @noetic-tools/core

@noetic-tools/core depends on @openrouter/agent (the OpenRouter LLM adapter) and zod; your package manager installs both automatically. Runs on Node.js 18+ and Bun.

Set your API key

Noetic talks to LLMs through OpenRouter. Grab a key at openrouter.ai/keys and export it — the runtime reads OPENROUTER_API_KEY by default:

export OPENROUTER_API_KEY=sk-or-...

Quick start

A complete, runnable program. Save it as agent.mjs and run node agent.mjs:

import { AgentHarness, step } from '@noetic-tools/core';

// 1. Define a single LLM step. `model` and `instructions` live here.
const greet = step.llm({
  id: 'greet',
  model: 'openai/gpt-4o-mini', // any OpenRouter model id (provider/model)
  instructions: 'You are a friendly assistant. Greet the user by name in one short sentence.',
});

// 2. Create the harness. `name` and `params` are required; `llm` selects the provider.
const harness = new AgentHarness({
  name: 'greeter',
  params: {},
  llm: { provider: 'openrouter' }, // apiKey defaults to process.env.OPENROUTER_API_KEY
});

// 3. Run the step with a context and print the model's reply.
const ctx = harness.createContext();
const reply = await harness.run(greet, 'Sam', ctx);
console.log(reply); // -> "Hello, Sam! How can I help you today?"

Using TypeScript? Save the file as agent.ts and run it with bunx tsx agent.ts (or bun run agent.ts) — both execute TypeScript directly, no build step.

Model ids use OpenRouter's provider/model format (e.g. openai/gpt-4o, anthropic/claude-sonnet-4) — see openrouter.ai/models.

Documentation

Full docs, specs, and examples live at noetic.tools/docs and in the main repo.

License

Apache-2.0