active
Memory Graph
Connects conversations, decisions, architecture, and retrieval ideas.
Memory Graph
The PUNNARAJ MUD should behave like a memory graph: each document is a node, and links show how intent, decisions, and implementation details depend on each other.
Current seed model
- Conversation context becomes Markdown notes.
- Notes are connected with wiki links.
- Search and graph navigation provide retrieval.
- Future AI context can use the vault as a retrieval source.
flowchart TD A[Conversation] --> B[Markdown note] B --> C[Linked MUD vault] C --> D[Static website] C --> E[AI retrieval context] E --> F[Better future updates]
Future vector layer
A future memory controller can index the Markdown vault into embeddings, but the canonical source should remain Markdown. This keeps the project portable and auditable.