How does Claude know when to use the stored memory?

Claude automatically checks for a MEMORY.md file at the start of every session and scans relevant entries before performing substantial tasks.

Does this work in monorepos?

Yes. In a monorepo setup, the skill uses the root of the specific sub-project you are working on to ensure the memory remains relevant to that specific context.

What happens if a stored memory conflicts with a new request?

If a memory entry conflicts with your current instructions, Claude is designed to surface the conflict and ask for clarification before proceeding.

Can I use Project Memory to store API keys?

No. The skill is strictly configured to never store secrets, tokens, credentials, or personal data to ensure your project's security.

Where is the project memory actually stored?

All project-specific data is stored in a file named MEMORY.md located at the root of your repository, making it easy to track via version control.

Project Memory

Name: Project Memory
Author: wunki

bywunki

•

Productivity & Workflow

Maintains a durable MEMORY.md file to track project conventions, user preferences, and past mistakes across AI sessions.

Project Memory transforms Claude from a stateless assistant into a context-aware collaborator by maintaining a persistent 'scratchpad' within your repository. By reading and writing to a standardized MEMORY.md file, the skill ensures that critical project knowledge—such as architectural decisions, specific coding styles, environment setups, and lessons learned from past errors—is preserved and applied automatically in future sessions. This eliminates repetitive prompting and helps the AI align more closely with your unique workflow and project requirements over time.

Key Features

0117 GitHub stars

02Categorized memory storage for preferences, conventions, mistakes, and environment notes

03Automated session initialization by loading existing project context

04Safe memory management with user-confirmed pruning of obsolete entries

05Natural language memory capture using triggers like 'remember this' or 'save to memory'

06Proactive lesson application that flags relevant memories before starting tasks

Use Cases

01Preventing the recurrence of specific bugs or logic errors by documenting past mistakes

02Onboarding the AI to unique local environment configurations and external service dependencies

03Enforcing project-specific naming conventions and architectural patterns without manual reminders

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add wunki/amplify project-memory

For use in Claude.ai and ChatGPT

Download Skill