Harlo

What it does

Harlo is an advanced AI coach that monitors behavioral patterns, predicts burnout, and manages cognitive state locally. It uses OpenUSD composition semantics to treat memory as a 3D-like spatial structure, ensuring total privacy and zero cloud dependency.

Tools

• coach: Primary interface for AI coaching and pattern analysis.
• store/recall: Low-latency memory storage and retrieval via SDR (Sparse Distributed Representation).
• query_past_experience: Semantic search over the user's cognitive history.
• patterns: Analysis of behavioral and cognitive rhythms.
• status: Health check of the current session and engine state.
• resolve_verifications: GVR (General Verification Routine) for ensuring state consistency.
• trigger_recalibration: Initiates a fresh cognitive intake calibration.

Installation

Install Harlo and its dependencies:

pip install -e .

For Claude Desktop, add the Harlo MCP server to your claude_desktop_config.json.

Supported hosts

• claude
• cursor