Documentation

Self-learning memory for AI agents — everything you need to get started

Hicortex gives your agents a shared long-term memory. They capture what they learn during the day, distill it into memories and lessons overnight, and surface the right thing at the right moment — automatically. Connect multiple agents and they learn together.

Quick start

  1. Single machine: npx @gamaze/hicortex init
  2. Connect more machines: npx @gamaze/hicortex init --server https://your-server:8787
  3. Done — capture, distillation, and recall run automatically. No cron setup.

Installation

Server mode, client mode, and the Hermes and OpenClaw plugins. Includes how the automatic capture and consolidation scheduling works.

Configuration

One-model LLM setup, scheduling, recall tuning, and the full config file reference.

Usage

How per-prompt recall, the memory dashboard, lessons injection, and the day-to-day pipeline work.

Domains

Define your knowledge spheres — Hicortex files every memory into them automatically and reorganizes nightly.

Upgrading

What changes when you move to 0.17: one-model config, the scheduling rework, and how your data is preserved.

API Reference

REST endpoints (including /distill and the analytics /dashboard) and the 9 MCP tools.

Changelog

What’s new in each release, starting with Version 0.17.