How does this skill save tokens?

By recalling distilled facts from a local SQLite database, Claude avoids having to read and process large source files to rediscover information it has already learned in previous sessions.

Can I use this with any repository?

Yes, this skill is designed to work with the claude_memory MCP server, allowing you to maintain a persistent knowledge base for any project you work on.

Why should I use /check-memory instead of just asking a question?

Using the explicit command ensures that Claude doesn't skip the retrieval step, which prevents it from hallucinating or ignoring established project conventions documented in previous sessions.

What is the check-memory skill for?

It forces Claude to check its persistent memory for project-specific knowledge and historical context before it starts searching code or providing answers.

What happens if the memory query returns no results?

The skill follows a fallback workflow: if memory is empty or partial, it proceeds to standard code exploration using tools like grep or read, clearly labeling what is new information.

Memory-First Workflow

Name: Memory-First Workflow
Author: codenamev

bycodenamev

•

生産性とワークフロー

Enforces a memory-first retrieval process to leverage persistent project knowledge before exploring code or answering queries.

The check-memory skill optimizes the Claude Code experience by mandating a knowledge-retrieval step before any file reading or code execution occurs. By querying a persistent SQLite-backed memory system via MCP tools, Claude can recall past architectural decisions, coding conventions, and previous lessons learned. This workflow significantly reduces redundant exploration and token usage while ensuring that every response is grounded in the historical context of the repository.

主な機能

01Source differentiation between recalled knowledge and new code discovery

02Token usage optimization through concise historical fact retrieval

03Mandatory memory-first verification to prevent redundant code exploration

04Support for conflict detection between current tasks and past implementation lessons

05Specialized shortcuts for recalling architecture, decisions, and conventions

068 GitHub stars

ユースケース

01Investigating the reasoning behind past technical decisions to avoid repeating mistakes

02Verifying specific project conventions before implementing a new feature to ensure consistency

03Onboarding to a codebase by retrieving previously stored architectural patterns

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add codenamev/claude_memory check-memory

For use in Claude.ai and ChatGPT

Download Skill