Can I filter my memory search by project?

Yes, all tools within this skill support project-based filtering to ensure you only retrieve history relevant to your current workspace.

What is the 3-layer workflow in Memory Search?

It is a token-saving strategy that starts with a lightweight search for IDs, expands context via a timeline tool, and finally fetches full details only for the necessary items.

Can I search for specific types of past work?

Yes, you can filter results by observation types such as bugfixes, features, decisions, discoveries, or general changes.

How does this skill help save tokens?

A search result uses only 50-100 tokens, whereas a full observation can use up to 1000. By filtering titles first, Claude avoids loading irrelevant content into its context window.

Memory Search for Claude

Name: Memory Search for Claude
Author: henriquescastilho

byhenriquescastilho

•

生産性とワークフロー

Searches and retrieves historical context from persistent cross-session memory to leverage past work and solutions.

Memory Search (mem-search) is a powerful retrieval skill that allows Claude to access a persistent database of previous interactions, decisions, and code changes across multiple sessions. By utilizing a specialized 3-layer workflow—Search, Timeline, and Batch Fetch—it enables Claude to efficiently recall how specific problems were solved in the past without wasting tokens on irrelevant data. This skill is essential for maintaining project continuity, avoiding redundant work, and providing Claude with long-term context that survives beyond a single chat window.

主な機能

011 GitHub stars

02Batch fetching capabilities to reduce API requests and latency

03Token-optimized 3-layer workflow (Search index, Context timeline, and Detailed fetch)

04Advanced filtering by project, observation type (bugfix, feature, decision), and date ranges

05Chronological timeline reconstruction to understand the sequence of past events

06Persistent cross-session memory retrieval for long-term project context

ユースケース

01Gathering historical context and facts to ensure consistency in ongoing development

02Checking if a specific feature or solution has already been implemented in a past session

03Recalling specific bug fixes or architectural decisions from previous weeks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add henriquescastilho/my-claude mem-search

For use in Claude.ai and ChatGPT

主な機能

011 GitHub stars

02Batch fetching capabilities to reduce API requests and latency

03Token-optimized 3-layer workflow (Search index, Context timeline, and Detailed fetch)

04Advanced filtering by project, observation type (bugfix, feature, decision), and date ranges

05Chronological timeline reconstruction to understand the sequence of past events

06Persistent cross-session memory retrieval for long-term project context

ユースケース

01Gathering historical context and facts to ensure consistency in ongoing development

02Checking if a specific feature or solution has already been implemented in a past session

03Recalling specific bug fixes or architectural decisions from previous weeks