Record and read Memory Bank state via mb-cli. Use for projects with memory-bank/database/ and a memory_bank.db. Triggers on: db workflow, record session work...
记忆
Mb Init
试用Initialize a memory bank for a new or existing project following the v6.12 protocol. Use when starting work on a project that lacks memory-bank documentation...
它能做什么
Initialize a memory bank for a new or existing project following the v6.12 protocol. Use when starting work on a project that lacks memory-bank documentation, or when creating a new project that needs task tracking. Creates the full memory-bank directory structure, initial files, and optionally a task file for the first piece of work.
技能文档
Memory Bank Initialization Skill (v6.12)
Overview
This skill bootstraps a complete memory-bank documentation system for any project. It creates the directory structure, initial files, and optionally a first task file — everything needed to start tracking work with the mb-text-workflow or mb-db-workflow skills.
When to Use
- Starting a new project that needs task tracking
- Adopting memory-bank documentation on an existing project that lacks it
- Creating a fork/adaptation of the memory-bank system for a specific project type
Prerequisites
- Determine the project root directory (absolute path)
- Determine the project name and purpose
- Decide if this project uses text-based (markdown) or DB-native workflow
- Know the current time and timezone
Workflow
Step 1: Create Directory Structure
cd
mkdir -p memory-bank/{tasks,sessions,edits/$(date +%Y-%m-%d),implementation-details}
Step 2: Create Core Files
memory-bank/tasks.md
# Memory Bank —
*Created: YYYY-MM-DD HH:MM:SS TZ*
*Last Updated: YYYY-MM-DD HH:MM:SS TZ*
## Overview
[Brief description of the project and what this memory bank tracks]
## Active Tasks
| ID | Title | Status | Priority | Started | Dependencies | Details |
|----|-------|--------|----------|---------|--------------|---------|
## Completed Tasks
| ID | Title | Status | Priority | Started | Completed | Dependencies | Details |
|----|-------|--------|----------|---------|-----------|--------------|---------|
## Status Summary
- **Active**: 0
- **Completed**: 0
- **Paused**: 0
- **Total**: 0
memory-bank/session_cache.md
# Session Cache
*Created: YYYY-MM-DD HH:MM:SS TZ*
*Last Updated: YYYY-MM-DD HH:MM:SS TZ*
## Current Session
**Started**: YYYY-MM-DD HH:MM:SS TZ
**Focus Task**: None
**Session File**: None
**Status**: 🔄 No active session
## Overview
- Active: 0 | Paused: 0 | Completed: 0
- Last Session: -
- Current Period:
## Task Registry
## Active Tasks
## Session History (Last 5)
## System Status
- **Memory Bank**: 🔄 Initialized, no tasks yet
- **Project**: 🔄 Ready for work
memory-bank/activeContext.md
# Active Context
*Last Updated: YYYY-MM-DD HH:MM:SS TZ*
## Current Tasks
## Completed Tasks (Recent)
## Next Steps
## System Status
memory-bank/edit_history.md
# Edit History
*Created: YYYY-MM-DD HH:MM:SS TZ*
*Last Updated: YYYY-MM-DD HH:MM:SS TZ*
---
## YYYY-MM-DD
#### HH:MM:SS TZ - INIT: Memory bank initialized
- Created `memory-bank/tasks.md` - Task registry
- Created `memory-bank/session_cache.md` - Session tracking
- Created `memory-bank/activeContext.md` - Current context
- Created `memory-bank/edit_history.md` - Edit history (this file)
- Created `memory-bank/implementation-details/` - Knowledge layer directory
Step 3: Create Optional Knowledge Layer Files
If the project has existing architecture or context worth capturing:
memory-bank/techContext.md— Stack, dependencies, build systemmemory-bank/productContext.md— Goals, user stories, featuresmemory-bank/systemPatterns.md— Conventions, patterns, standardsmemory-bank/changelog.md— Feature/release trackingmemory-bank/errorLog.md— Known issues and fixes
Step 4: Create First Task (Optional)
If there's immediate work to track:
# Create task file
cat > memory-bank/tasks/T1.md << 'EOF'
# T1: [Title]
*Created: YYYY-MM-DD HH:MM:SS TZ*
*Last Updated: YYYY-MM-DD HH:MM:SS TZ*
**Status**: 🔄 **IN PROGRESS**
**Priority**: HIGH
## Details
[Brief description]
## Progress
1. 🔄 [Current item]
2. ⬜ [Next item]
## Files
EOF
# Update tasks.md registry
# (prepend T1 to Active Tasks table)
Step 5: Commit Initialization
git add memory-bank/
git commit -m "(docs)INIT: Initialize memory bank — v6.12 protocol, task tracking ready"
Post-Initialization
After initialization, use mb-text-workflow or mb-db-workflow for all subsequent updates:
- Create edit chunks in
memory-bank/edits/YYYY-MM-DD/ - Update task files in
memory-bank/tasks/ - Regenerate
memory-bank/edit_history.md - Commit with format:
(type)TID: Headline - Details
Project-Specific Adaptations
For different project types, customize:
Research projects:
- Add
papers/directory in implementation-details - Track literature review progress in tasks
Web/apps:
- Add
api-endpoints.md,ui-components.md - Track feature phases (MVP, v2, etc.)
DevOps/infrastructure:
- Add
deployment.md,monitoring.md - Track environment parity
Anti-Patterns
- NEVER initialize a memory bank inside another project's memory-bank directory
- NEVER use relative paths or tildes in file references
- NEVER skip the edit_history.md — it's the audit trail
- NEVER commit generated views (edit_history.md) without their source chunks
References
- mb-text-workflow — For text-based projects
- mb-db-workflow — For DB-native projects
- v6.12 Full Rules
相关技能
Memory Bank text-based update workflow following integrated-rules v6.12. Use when updating memory bank files manually via markdown editing — creating edit chunks, updating tasks.md, session_cache.md, session files, and task files. NOT for database-native workflows (use mb-db-workflow instead). Triggers on phrases like "update memory bank", "create edit chunk", "update tasks.md", "update session cache", "memory bank update workflow".
Memory Bank Update Workflow — v6.12 compliant with project-repo awareness. Use when the user invokes $mem-update or asks for this workflow by name.
Global workflow for loading memory bank files based on recent activity and context Use when the user invokes $mem-load or asks for this workflow by name.
Memory Bank Template Compliance Cleaner Workflow Use when the user invokes $mem-format or asks for this workflow by name.
A cross-agent project memory skill for Codex and Claude Code, enabling consistent context, documentation, and development history across sessions.