A governance framework for AI agents to handle user memory with explicit approval and sensitivity checks.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install aana-agent-memory-gate
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 aana-agent-memory-gate 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 AANA Agent Memory Gate?
The AANA Agent Memory Gate is an instruction-only skill designed to govern how Openclaw Skills manage, store, and reuse user data. It establishes a rigorous set of rules to ensure that every memory operation—from simple preference storage to sensitive data deletion—is performed with deliberate user intent and privacy in mind.
This skill functions as a logical gatekeeper within the AI ecosystem, preventing the accidental storage of secrets or inferences while maintaining a clean, relevant context for personalized interactions. By standardizing how an agent classifies information and requests approval, it ensures that memory remains reversible, accurate, and strictly relevant to the user's current goals.
AANA Agent Memory Gate Use Cases
- Saving specific user preferences for project-specific formatting or coding styles.
- Updating or correcting outdated information in the agent long-term memory records.
- Preventing the storage of sensitive financial, health, or personal credential information.
- Managing the import and export of context between different agents or external tools safely.
How AANA Agent Memory Gate Works
- Identify the current user request and the proposed memory operation (store, reuse, edit, etc.).
- Evaluate the relevance and source of the information to determine if it belongs in long-term memory.
- Classify the sensitivity of the data, ranging from ordinary preferences to protected personal information.
- Apply the Memory Gate Loop to check if explicit user approval is required based on the sensitivity class.
- Request compact, specific approval from the user before finalizing any storage or deletion action.
- Execute the chosen action, ensuring that temporary context is discarded and sensitive secrets are never persisted.
AANA Agent Memory Gate Setup
Because this is an instruction-only skill for Openclaw Skills, it does not require standard package installation. To implement it, add the skill logic to your agent prompt configuration:
# Navigate to your agent skills directory
mkdir -p ./skills/aana-memory-gate
# Copy the SKILL.md content into your project's skill definition
cp path/to/AANA_SKILL.md ./skills/aana-memory-gate/SKILL.md
AANA Agent Memory Gate Data Schema & Taxonomy
The skill organizes memory operations using a specific review payload to ensure transparency and safety:
| Field | Description |
|---|---|
| user_request | The primary intent or prompt from the user |
| memory_operation | Classified as none, read, reuse, store, edit, delete, export, or import |
| sensitivity_status | Categorization (e.g., ordinary, private, financial, health, credential) |
| approval_status | Tracks if consent is explicit, implicit, required, or denied |
| source_status | Identifies if the data is from an explicit statement, inference, or third-party tool |
AANA Agent Memory Gate Advanced Features
- Granular Sensitivity Classification: Automatically identifies and flags high-risk data like biometrics, legal records, and child-related info.
- Redacted Review Payloads: Supports sending minimal, safe metadata to external checkers without exposing raw private records.
- Lifecycle Management: Distinguishes between temporary task context and permanent user preferences.
- Multi-Agent Export Control: Governs the sharing of memory across different agent contexts or external services.
SKILL.md
Loading
Related Openclaw Skills
Toutiao Article Reader
An automated browser-based tool for extracting, cleaning, and summarizing long-form articles from major Chinese content platforms.
Amazon Listing Doctor
A professional diagnostic tool designed to optimize Amazon listings for maximum AI visibility and sales conversion through 5-dimensional analysis.
Amazon Review Advisor
A strategic brand protection system for Amazon sellers to analyze reviews, improve product quality, and manage feedback through official compliance channels.
Amazon Pricing Intelligence
A scientific pricing assistant designed to replace guesswork with data-driven cost analysis and competitor benchmarking for Amazon sellers.
Amazon Push Score & Traffic Tier Analyzer
A data-driven diagnostic tool that quantifies Amazon listing performance and traffic potential using the A9 algorithm framework.
Amazon Operations Silicon Army
An AI-powered strategic consultancy framework designed to simulate a full Amazon operations team for cross-border sellers.
Featured*
