OCD FAQs

Question 1

How does OCD benefit LLMs and AI applications?

Accepted Answer

OCD gives LLMs persistent, structured memory across sessions, preventing context loss. It offers token-efficient context retrieval using a `tree-text` format, fetching only necessary branches, and supports advanced hierarchical pattern queries for precise information access.

Question 2

What is 'token-efficient context retrieval' in OCD?

Accepted Answer

OCD's token-efficient context retrieval means that when an LLM requests information, OCD fetches only the specific relevant branches of your hierarchical knowledge using the `tree-text` format, rather than entire, potentially large documents. This significantly reduces token usage and improves performance.

Question 3

What is OCD (Organized Context Datastore)?

Accepted Answer

OCD is an MCP (Model Context Protocol) server designed to provide LLMs and humans with a collaborative, Git-backed datastore. It stores project knowledge as hierarchical Markdown files, ensuring a shared, persistent source of truth for AI and human collaboration.

Question 4

How do humans interact with the knowledge stored in OCD?

Accepted Answer

Humans can easily interact with OCD's knowledge base using plain Markdown files with frontmatter. You can edit content in any text editor (like VS Code), review changes on GitHub, or use the built-in Web UI for browsing and editing, all while sharing the same Git-backed source.

Question 5

Is OCD truly Git-backed, and what are the benefits?

Accepted Answer

Yes, OCD is fully Git-backed. All project knowledge is stored in a Git repository, providing version control, commit history, and the ability to revert changes. This ensures data persistence, collaborative review workflows, and a reliable audit trail for both AI-generated and human-edited content.

OCD

OCD

主な機能

ユースケース

主な機能

ユースケース