A sophisticated memory layer that tracks user preferences, goals, and boundaries to provide highly personalized AI interactions over time.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install remember-me
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 remember-me 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 Remember Me?
The Remember Me skill provides a robust framework for AI agents to maintain a respectful and actionable model of the user. Unlike simple conversation logs, this skill categorizes information into facts, preferences, and goals, ensuring that every Openclaw Skills interaction becomes increasingly tailored to the user's specific workflow. By differentiating between explicit facts and inferred hypotheses, the agent can adapt its behavior—such as tone, brevity, and decision-making style—without overstepping or making incorrect assumptions.
This system is designed to evolve alongside the user, promoting short-term observations into a durable long-term profile. Whether it is remembering architectural boundaries in a coding project or adhering to specific communication styles, the skill ensures that Openclaw Skills remain aligned with user expectations without requiring constant re-instruction.
Remember Me Use Cases
- Storing technical preferences and architectural constraints for software development projects.
- Tracking long-term project goals and deadlines to provide proactive assistance.
- Maintaining a list of behavioral boundaries, such as "don't do X" or "keep Y private."
- Recording recurring blockers or patterns that emerge during daily workflows.
- Personalizing the agent's interaction style based on historical success and user feedback within Openclaw Skills.
How Remember Me Works
- The agent identifies memory triggers, such as explicit "remember this" requests or recurring behavioral patterns.
- Signals are classified into four types: FACT, PREFERENCE, GOAL, or HYPOTHESIS.
- Raw events are appended to a daily note file (
memory/YYYY-MM-DD.md) with a timestamp. - Inferred patterns are stored as hypotheses with a confidence score (high, medium, or low).
- The agent validates hypotheses through lightweight conversational check-ins during natural interactions.
- Validated or high-impact items are promoted to the long-term
MEMORY.mdfile based on a structured promotion workflow.
Remember Me Setup
To enable persistent memory in your Openclaw Skills environment, ensure your workspace includes the following directory structure:
mkdir -p memory
mkdir -p references
touch memory/MEMORY.md
You must also define the following reference files to guide the agent's classification logic:
references/templates.md: Templates for memory entries.references/promotion-checklist.md: Criteria for moving data to long-term memory.references/profile-schema.md: The structure for the durable user profile.
Remember Me Data Schema & Taxonomy
The skill utilizes a tiered storage approach to balance granularity with long-term utility:
| Component | File Path | Purpose |
|---|---|---|
| Daily Notes | memory/YYYY-MM-DD.md |
Stores raw, timestamped events and low-confidence hypotheses. |
| Long-term Memory | MEMORY.md |
A curated profile containing durable preferences, goals, and boundaries. |
| Reference Metadata | references/ |
Contains the schemas and checklists that maintain memory integrity. |
Every entry is tagged with its integrity status (e.g., FACT vs HYPOTHESIS) and an optional Memory Impact Score (1-3) to determine its priority in the promotion workflow.
Remember Me Advanced Features
- Confidence Decay: Automatically downgrades unreinforced hypotheses over 14, 30, and 60-day periods to keep memory fresh.
- Memory Impact Scoring: Heuristics that prioritize workflow-affecting and outcome-critical information for immediate promotion.
- Assumption Loop: A continuous logic cycle where the agent observes behavior, forms hypotheses, and tests them with non-intrusive probes.
- Forgetting & Demotion Policy: Explicit rules for removing contradicted preferences or stale project context from Openclaw Skills.
- Weekly Maintenance: Recommended review cycles to merge stable patterns and remove redundant or speculative entries.
SKILL.md
Loading
Related Openclaw Skills
Omarchy
Omarchy provides a comprehensive suite of system wrappers and operating guardrails to ensure AI agents manage local environments using safe, native workflows.
Glance
Glance is an AI-extensible personal dashboard that allows developers to create custom widgets using natural language, where AI agents handle the underlying data collection and caching.
Redshift CLI for Decentralized Secret Management
Redshift is a decentralized secret management tool that uses Nostr relays for client-side encrypted storage, eliminating the need for central servers.
PPT Compression
A specialized tool for drastically reducing PowerPoint file sizes by optimizing internal media and converting the output to compact PDF format.
Soaring Weather
A professional-grade thermal and gliding weather forecast tool using an 11-parameter scoring engine to predict soaring conditions.
Humanizer AI Text Refinement
A sophisticated text processing skill that identifies and removes AI-generated signatures to transform robotic LLM output into natural human writing.
METADATA
Featured*
