A context and memory system for AI coding agents, this project offers a reference implementation demonstrating persistent memory, personal rules, skills, and scheduled observations. It targets developers and researchers aiming to build more adaptive and context-aware AI assistants by providing a blueprint for system architecture, data flow, and memory accumulation. The primary benefit is enabling users to understand and replicate advanced AI agent capabilities.

How It Works

The system employs a layered design, distinguishing between reference examples and directly usable components. It features a three-tier memory system (observations, daily records, thought reviews) and organizes agent behavior through rules, axioms, and skills. Core components include scheduled jobs for daily observation and weekly reflection, alongside tools for semantic search and report sharing. The architecture emphasizes data flow and memory accumulation, serving as a blueprint rather than a plug-and-play solution.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/grapeot/context-infrastructure) and navigate into the directory.
• Environment: Open the project in an AI-aware IDE like Claude Code, OpenCode, or Cursor.
• Configuration: Edit rules/USER.md to input personal preferences and background information for AI personalization. Detailed setup instructions are available in setup_guide.md.
• Prerequisites: No specific non-default prerequisites are listed beyond standard Python development environments and an AI-aware IDE. Cron jobs are required for periodic memory updates.

Highlighted Details

• Reference Implementation: Provides a complete structure and data flow for a year-long context infrastructure system.
• Layered Design: Differentiates between display-layer examples (axioms, skills) and reusable components (user rules, communication styles, memory system code).
• Personalization Focus: Emphasizes user data collection and configuration (rules/USER.md, rules/SOUL.md) as crucial for tailoring AI behavior.
• Memory System: Includes scheduled observer.py and reflector.py scripts for daily and weekly memory updates, documented in periodic_jobs/ai_heartbeat/docs/.

Maintenance & Community

The README does not provide specific details on notable contributors, sponsorships, community channels (like Discord/Slack), or a public roadmap. It focuses primarily on the technical structure and usage as a reference.

Licensing & Compatibility

The project is released under the MIT License. This permissive license generally allows for commercial use, modification, and distribution, with minimal restrictions beyond attribution.

Limitations & Caveats

This project is explicitly positioned as a reference implementation and blueprint, not an out-of-the-box tool. Achieving true AI personalization requires users to collect and integrate their own behavioral data, as the provided axioms and skills are examples derived from the original author's experience and are not directly transferable without adaptation. Setup involves significant configuration and data input.