Lite SuiteLite Suite
← Back to Docs

LiteAgent

Autonomous AI agent with identity, memory, and heartbeat scheduling

Overview

LiteAgent is an autonomous AI agent desktop app. It has identity files (soul, user, memory), heartbeat scheduling, an embedded browser, a Python CLI agent loop, DAG-based memory (LCM), marketing skills, and local LLM support via LM Studio. It remembers, it acts, it learns.

Key Features

  • Identity System — Soul, user, and memory files give the agent a persistent sense of self across sessions
  • Heartbeat Scheduling — Configurable intervals for autonomous operation. The agent wakes, works, and sleeps on schedule
  • Python CLI Agent — Interactive REPL with auto-compaction, MCP tool access, and permission management
  • Embedded Browser — Built-in Chromium for web research and automation
  • Dual-Layer Memory — Local SQLite (importance-scored) + LCM DAG (hierarchical summaries)
  • Marketing Skills — 13 built-in skills for content, analytics, social media, and ad creative
  • Claude Code Integration — Command Center panel with live status for loops, schedules, connectors, and dispatch

Architecture

| Port | Service | Auth | |------|---------|------| | 7430 | Agent API Server | None (localhost-only) | | 7431 | Agent Bridge | Bearer token |

Heartbeat Service

The heartbeat is the autonomous core. On each tick:

  1. Assembles identity context from soul/user/memory/heartbeat files
  2. Injects LCM DAG context (summaries + recent messages)
  3. Spawns either claude CLI or liteagent agent (LM Studio)
  4. Stores run output as LCM messages
  5. Configurable: interval, max turns, timeout, notification mode

Python CLI

liteagent agent        # Interactive agent REPL
liteagent browser      # Browser automation
liteagent content      # Content generation
liteagent analytics    # Analytics queries
liteagent identity     # Identity management
liteagent skill        # Skill CRUD