search-memory

Search memory episodes semantically and rank results by relevance and recency.

Updated Jan 24, 2026
One-click install
npx skills add https://github.com/yoshiwatanabe/yoshiwatanabe-plugins --skill search-memory-yoshiwatanabe
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: search-memory
Source: https://github.com/yoshiwatanabe/yoshiwatanabe-plugins/tree/main/dev-memory/skills/search-memory
Command: npx skills add https://github.com/yoshiwatanabe/yoshiwatanabe-plugins --skill search-memory-yoshiwatanabe

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill enables developers to quickly locate past memory episodes by understanding intent rather than relying on exact keyword matches, saving time during debugging and project recall.

Core Features & Use Cases

  • Semantic search of development memory episodes to surface relevant entries beyond keyword matching
  • Ranking by relevance and recency to surface the most useful memories
  • Use Case: When trying to recall a debugging session or a decision from a previous day, query in natural language to retrieve the best-matching episode metadata and summary

Quick Start

Use the search-memory skill with a natural language query like "What did I work on regarding memory management last week?" and set an optional limit to control how many episodes are returned.

Frequently Asked Questions about search-memory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search through development memory episodes using natural language instead of exact keywords?

Semantic search of development memory episodes interprets natural language intent to surface relevant entries beyond exact keyword matching, reading frontmatter from your config repository to rank and return the most useful past debugging or project memories.

What is semantic search for development memory and how does it rank past episodes?

Semantic search for development memory locates past episodes by understanding query intent, ranking results by relevance and recency. It enumerates stored files using Glob, reads episode frontmatter for matching, and sorts returns by filename with the most recent first.

Do I need a specific environment variable to locate memory episodes in my configuration repository?

Yes, locating memory episodes requires the YW_CONFIG_REPO_PATH environment variable to find the configuration repository. Episodes must be stored under the domains/dev/memory/episodes directory structure for the search tooling to enumerate and read them.

Can I limit the number of memory episodes returned by a semantic search query?

Yes, you can set a user-defined limit to control how many memory episodes the semantic search returns. Results are sorted by filename with the most recent first, ensuring the most relevant and recent development memories surface within your specified limit.

What's the best way to recall a past debugging session or technical decision from previous days?

The best way to recall past debugging sessions or technical decisions is querying your development memory episodes with natural language. Semantic search surfaces relevant episode metadata and summaries by understanding intent, saving time during project recall beyond exact keyword matching.