Memory

To organize long-term structured data, you can use a categorized local memory system with a root directory, configuration file, and per-category index files to store unlimited projects, people, and decisions. This runs alongside built-in memory without replacing it.

The best way to set up a local knowledge management directory is to create a ~/memory/ workspace, add a config.md file for settings, and create an INDEX.md at the root level alongside category folders to maintain discoverability.

Yes, you can sync built-in agent memory with a custom structured storage system. The setup supports optional synchronization from built-in memory into your categorized folders, allowing you to expand and resync data as it grows.

Indexing for unlimited categorized memory collections works by maintaining a root INDEX.md for global discovery and individual INDEX.md files within each category folder, ensuring all stored knowledge, projects, and people remain easily discoverable.

No external dependencies are required to run a persistent memory system. You only need a local ~/memory/ directory to hold your configuration, category folders, and index files, with optional references and scripts for extended functionality.

You should avoid using a file-based memory system for knowledge management if your data requires real-time multi-user concurrent editing or complex relational queries, as this system relies on local markdown files and directory structures for persistence.