skills.rest
a2mus
a2musCOMMUNITY
Explore all skills by this author →
GitHub

memory-bank

Initialize, read, update, and review structured project memory files across sessions.

Updated Apr 7, 2026
Search Tags:#git#templates#agent#memory#persistence#session#handover
One-click install
npx skills add https://github.com/a2mus/smart-da3m --skill memory-bank-a2mus
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-bank
Source: https://github.com/a2mus/smart-da3m/tree/main/.agent/skills/memory-bank
Command: npx skills add https://github.com/a2mus/smart-da3m --skill memory-bank-a2mus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Provides a structured, persistent project memory so agents and collaborators can resume work across sessions without losing context, decisions, or progress.

Core Features & Use Cases

  • Ordered Boot Sequence: mandates reading core memory files in a specific order at session start to establish the current state and context.
  • File Templates & Safe Edits: includes templates for current-state, project brief, product and tech context, patterns, progress, and next-session with rules to read-before-write and append-only updates.
  • Operational Rules & Handover: defines initialization and closing protocols, date verification, gardening rules for architecture decisions, and guidance to prevent accidental deletion or restructuring.
  • Use Case: ideal for agent-driven development workflows where continuity, reproducible handovers, and an auditable project history are required.

Quick Start

Initialize the memory bank by creating memory-bank/core/, generate the seven core files from the provided templates, and load them in order from current-state.md through next-session.md.

Frequently Asked Questions about memory-bank

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I maintain agent context and project state across sessions?▼

To maintain agent context across sessions, use a persistent memory store that initializes structured project memory files during session boot and updates them on closing. This ensures consistent handovers and preserves progress without losing state.

What is the best way to implement an agent handover protocol for development continuity?▼

An effective agent handover protocol uses an ordered boot sequence to read core memory files like current-state and next-session at startup, then appends progress updates before closing to ensure the next session resumes with full context.

How do I prevent data loss when updating persistent project memory files?▼

To prevent data loss when updating persistent project memory, enforce read-before-write safeguards and append-only updates. These operational rules preserve history and prevent accidental deletion or restructuring of context files.

Can I use structured memory files for agent-driven workflows without external dependencies?▼

Yes, you can use structured memory files for agent-driven workflows without external dependencies. The system operates using file templates and local directories, requiring no additional packages to manage project context and continuity.

What limitations exist when using file templates for session memory persistence?▼

Limitations of file-based session memory persistence include the strict adherence required to the ordered boot sequence and gardening rules; failing to follow initialization protocols risks breaking context continuity or losing architectural decisions.

Related Skills

Explore More Software Engineering Skills→
artsmcartsmc
3

documentation-start

Initialize missing Memory Bank and Document Hub documentation structures.

Community
Basic
saolalabsaolalab
37

memory

Store long-term facts in MEMORY.md and searchable HISTORY.md event logs.

Official
Intermediate
h4vzzh4vzz
33

version-control

Manage Git repositories and collaborative workflows across GitHub, GitLab, and Bitbucket.

Community
Advanced
spm1001spm1001
3

beads

Persist cross-session project memory using graph-based dependency models.

Community
Advanced