# Initialize Local AI Memory

AI assistants often lose project context between sessions. This skill creates local memory files and clear routines for Claude, Codex, and Claude Code.

## Install

```bash
npx skillstore add yunshu0909/memory-init
```

## Metadata

- Status: approved
- Slug: yunshu0909-memory-init
- Skillstore revision: r1
- Version status: missing
- Tree hash: 40bba4d042281b4c0f18787d14c070818f8a9cc328831d9f0b1546669b4394af
- Author: yunshu0909
- GitHub username: yunshu0909
- License: MIT
- Repository: https://github.com/yunshu0909/yunshu\_skillshub/tree/master/memory-init
- Ref: 64ca8af0f54a325752f08bd54e52151061ea659a
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, filesystem
- Quality score: 77
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/yunshu0909-memory-init
- Manifest: https://skillstore.pages.dev/api/skills/yunshu0909-memory-init/manifest

## Capabilities

- Checks for existing CLAUDE.md, AGENT.md, MEMORY.md, and memory/ before initializing.
- Collects project purpose, user role, assistant use, style preferences, and work habits.
- Generates CLAUDE.md with project context, persona, conversation style, thinking priorities, and memory protocol.
- Optionally creates AGENT.md with the same instructions for Codex.
- Creates an initial MEMORY.md template for long-term facts and decisions.
- Creates a memory/ directory and can add a first daily note with user approval.

## Use Cases

- Start Project Memory: Create local memory files before using Claude Code or Codex in a new repository.
- Standardize Context: Give assistants a consistent project profile, user role, style preference, and memory workflow.
- Capture Daily Decisions: Record useful decisions and context in daily memory notes during ongoing work.

## Prompt Templates

### Initialize Memory

```
Run /memory-init in this project. Ask me the setup questions and create the memory files after I answer.
```

### Add Codex Support

```
Set up project memory and include AGENT.md for Codex if I confirm the generated CLAUDE.md content.
```

### Protect Existing Files

```
Inspect this directory for existing memory files. Tell me what exists before adding missing files or overwriting anything.
```

### Build a Tailored Memory Workflow

```
Initialize memory for this project. Infer the project summary from files, then tailor persona, style, priorities, and daily memory rules.
```

## Limitations

- Works only with files in the current project directory.
- Does not synchronize memory across devices or repositories.
- Generated memory files may contain sensitive details if users provide them.
- Existing files require careful review before overwrite or reset.

## Best Practices

- Review generated CLAUDE.md and AGENT.md before relying on them.
- Keep MEMORY.md focused on stable facts, decisions, and preferences.
- Avoid storing secrets, credentials, or private customer data in memory files.

## Anti Patterns

- Do not overwrite existing assistant instruction files without review.
- Do not use daily memory notes as a full task tracker.
- Do not store sensitive data that should stay outside repository files.

## Security Audit

- Audited at: 2026-07-08T10:48:02.185\+00:00
- Summary: The static alerts are false positives caused by Markdown backticks, fenced examples, and references to local memory files. The references to ~/.claude/ explicitly instruct the assistant not to read or write the global directory. No prompt injection attempt, data exfiltration intent, or hidden global-file access was found in SKILL.md.

## Stats

- Views: 166
- Downloads: 27
- Favorites: 3
- Popularity score: 0