memory-search

To search an agent memory base without exact file paths, you can use frontmatter fields, full-text keywords, or cross-reference graph traversal to retrieve a ranked list of relevant file paths and excerpts.

Cross-reference graph traversal for information retrieval is a search mechanism that discovers linked files by following related decision IDs up to depth one, returning relevant file paths and excerpts based on their connections.

Yes, you can find knowledge base files using structured YAML frontmatter fields. The search supports filtering by metadata like domain, tags, or related decisions to locate relevant memory files.

To perform full-text keyword search across memory files, input specific terms into the content search function. It scans all files and returns a ranked list of IDs, titles, and excerpts prioritizing relevance.

Yes, memory search returns results that stay within token limits. It prioritizes relevance and provides a constrained list of file paths, IDs, titles, and excerpts to ensure efficient information retrieval.

The best way to retrieve unknown decision documents about topics like pool exhaustion is using full-text keyword or cross-reference search to instantly locate and rank relevant memory files without knowing exact locations.