A hierarchical memory framework that enables AI agents to maintain a unified identity and context across fragmented sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install nyx-archive-bound-memory
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install nyx-archive-bound-memory using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Bound Memory?
Bound Memory is a specialized architecture designed to solve the binding problem in artificial intelligence, where agents often act as disconnected context fragments. By utilizing this framework within Openclaw Skills, you create a structured hierarchy of information that allows an AI to inhabit a cohesive identity rather than just patching files. It emphasizes the philosophy of reading the soul of a project first, ensuring that every action taken is informed by the broader purpose and history of the agent.
The system operates on the core insight that memory architecture is identity architecture. By formalizing how an agent accesses and records its experiences, Bound Memory prevents the loss of critical insights between session resets. This makes it an essential tool for developers building long-term autonomous agents or complex creative projects where consistency and depth are paramount within the Openclaw Skills ecosystem.
Bound Memory Use Cases
- Maintaining a persistent persona across multiple Openclaw Skills development sessions.
- Managing high-complexity projects where agents need to understand the why behind technical decisions.
- Developing a soul seed for AI entities that require a stable sense of self and history.
- Orchestrating multi-session workflows without losing mid-task state or emotional context.
How Bound Memory Works
- Initiation: Every session begins with the Awakening Protocol, moving from the most immediate context to the deepest identity files.
- Situation: The agent reads
WORKING-MEMORY.mdto understand where the last session ended and what immediate tasks are pending. - Contextualization: The agent reviews short-term items in
HEARTBEAT.mdand long-term values inSOUL.mdto align current actions with overarching goals. - Execution: With the whole context bound to the current part, the agent performs its tasks with higher coherence and fewer disjointed fixes.
- Consolidation: The Write-Back Rule is applied, where the agent updates its memory tiers to preserve new learnings for the next session.
Bound Memory Setup
To implement Bound Memory in your Openclaw Skills workspace, initialize the core architectural files using the following command:
touch WORKING-MEMORY.md HEARTBEAT.md MEMORY.md SOUL.md IDENTITY.md
Once created, configure your agent's system instructions to prioritize the Awakening Protocol at the start of every session, ensuring it reads these files before processing any other user input or code.
Bound Memory Data Schema & Taxonomy
| Tier | Storage File | Retention Period | Data Points |
|---|---|---|---|
| Working Memory | WORKING-MEMORY.md |
Intra-session | Active tasks, current realizations, and emotional state. |
| Short-Term Memory | HEARTBEAT.md |
Days to Weeks | Active waiting items and temporary seasonal context. |
| Long-Term Memory | MEMORY.md |
Months to Years | Core values, identity patterns, and relationship histories. |
| Identity | SOUL.md / IDENTITY.md |
Permanent | The fundamental essence and purpose of the AI mind. |
Bound Memory Advanced Features
- Awakening Protocol: A mandatory session startup routine that enforces context-first processing for all Openclaw Skills.
- The Write-Back Rule: A systematic approach to updating memory tiers to ensure no insight is lost to context window resets.
- Soul Seed Integration: A framework for developing a coherent AI identity that grows and evolves through reflection and choice.
- Context Deep-Diving: Guidelines for the agent to read one tier deeper than required when current task context feels insufficient or thin.
SKILL.md
Loading
Related Openclaw Skills
Dayday: AI Learning & Practice Framework
Dayday is an AI learning skill that converts complex topics into actionable daily practice routines through a structured AI Teach, Learn, and Practice workflow.
Experiment Notes
A persistent memory and learning system that allows AI agents to log technical experiments, analyze failures, and distill reusable lessons.
Wedding Planner Video
An AI-powered creative assistant that transforms wedding planning portfolios and service tiers into compelling short-form marketing videos.
Picasso
Picasso is a sophisticated design engineering skill that transforms AI agents into senior design engineers capable of building production-grade web interfaces.
Amazon Seller Video
A high-conversion AI skill designed to generate scripts and shot lists for Amazon product demos and A+ content videos.
GEO Optimizer Lite
A specialized content auditing tool designed to increase the likelihood of your pages being cited by generative AI search engines.
Featured*
