An automated diagnostic tool that identifies abnormal metric fluctuations by comparing real-time data against historical statistical baselines.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install aloudata-anomaly-detection
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 aloudata-anomaly-detection 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 Anomaly Detection?
The Anomaly Detection skill is a specialized component within the Openclaw Skills library designed to evaluate the health of business metrics. Unlike simple query tools that only return values, this skill applies logic to determine if data is normal or deviant. It acts as the middle layer of an analysis pipeline: first fetching data via metric queries, then identifying problems through detection, and finally suggesting attribution for root cause analysis.
By leveraging statistical models like standard deviation (sigma) or business-defined thresholds, the skill provides a structured assessment of data points. This allows developers and analysts to move beyond manual monitoring and adopt an exception-based management style where only significant deviations require human intervention.
Anomaly Detection Use Cases
- Performing daily health inspections on core KPIs to ensure business stability.
- Identifying sudden sales fluctuations across specific channels or brands that might be masked by total aggregates.
- Monitoring inventory levels for stockout risks or identifying slow-moving items in real-time.
- Automated scanning of high-cardinality data to flag outliers that deviate from 30-day historical patterns.
How Anomaly Detection Works
- Scope Definition: The skill identifies which metrics and dimensions (e.g., channel, region) need to be inspected based on user intent.
- Baseline Establishment: It calculates what constitutes normal behavior using historical trends, period-over-period growth, or statistical standard deviations.
- Data Retrieval: The skill delegates the actual data fetching to a metric-query tool to ensure up-to-date information is used.
- Logical Analysis: It executes a detection algorithm (e.g., Python-based sigma analysis) to compare current values against the established baseline.
- Reporting: A structured report is generated, categorizing results into Critical, Warning, or Normal status.
- Attribution Hand-off: If a significant anomaly is detected, it recommends moving to the attribution phase to find the root cause.
Anomaly Detection Setup
To use this skill within the Openclaw Skills environment, ensure you have the necessary query dependencies installed. This skill specifically requires the metric-query skill to handle data ingestion.
# Install the core skill package
openclaw install anomaly-detection
# Ensure Python environment is ready for statistical calculations
pip install numpy
Anomaly Detection Data Schema & Taxonomy
The skill organizes its output into a structured format to facilitate both human reading and machine processing. Below is the primary metadata taxonomy used in reports:
| Property | Description | Format |
|---|---|---|
| metric_name | The identifier of the KPI being monitored | String |
| dimension_value | The specific segment (e.g., 'Retail' or 'Wholesale') | String |
| current_value | The actual observed value in the current period | Numeric |
| baseline_value | The calculated normal value or threshold | Numeric |
| deviation_sigma | The number of standard deviations from the mean | Float |
| severity_level | Categorization: Normal (Check), Warning (Caution), or Critical (Alert) | Enum |
Anomaly Detection Advanced Features
- Dynamic Statistical Baselines: Automatically calculates mean and standard deviation from 30-day windows to adjust for varying metric volatility.
- Multi-Dimensional Drill-down: Scans across sub-dimensions to find hidden anomalies that do not show up in top-level totals.
- Trend-Pattern Recognition: Detects continuous decline or growth patterns over multiple days rather than just single-point deviations.
- Custom Threshold Overrides: Allows power users to define absolute business red-lines (e.g., 'alert if inventory < 500') that override statistical models.
- Seamless Integration: Designed to work within an agentic workflow, passing state to attribution skills for immediate problem-solving.
SKILL.md
Loading
Related Openclaw Skills
Moore Pyramid Memory System
A sophisticated 5-layer memory architecture that provides AI agents with persistent, cross-session continuity and automated task tracking.
Agency Orchestrator
An intelligent multi-agent coordination system that automatically analyzes tasks and selects the best AI agents to collaborate on complex projects.
AVIC Information Assistant
A specialized assistant for querying products, news, financial data, and official resources for the Aviation Industry Corporation of China.
Chargeback Assistant
An automated expert system designed to help consumers navigate complex payment network reason codes and successfully file transaction disputes.
Arsenal FC
A dedicated AI agent skill for fetching real-time Arsenal FC statistics, news, and historical data.
Avatr Information Assistant
A specialized research skill for retrieving real-time product data, financial reports, and official news regarding the Avatr Smart EV brand.
Featured*
