scan

What This Skill Does

Scan is the core sensory module for OpenClaw agents, serving as the Universal Perceptual Interface. It transcends simple data retrieval by performing high-density information extraction across multiple modalities. Whether your agent is tasked with parsing complex genomic VCF files, conducting deep-packet forensic analysis, or mapping a physical room via LiDAR data, Scan provides the bridge between raw, unstructured reality and actionable semantic vectors. By leveraging the SCAN_MATRIX, the skill categorizes incoming data into four primary domains: biometric, forensic, semantic, and spatial, ensuring that the agent applies the correct heuristic analysis for every input type.

Installation

To integrate the Scan skill into your existing agent environment, execute the following command via the OpenClaw terminal:

clawhub install openclaw/skills/skills/agenticio/scan

Ensure your local environment has the necessary dependencies for heavy-duty data processing, particularly if you intend to use the spatial mapping or genomic analysis components.

Use Cases

Scan is designed for high-stakes, information-dense environments. Common use cases include:

• Litigation & Compliance: Rapidly digesting thousands of pages of legal dockets to highlight specific evidentiary anomalies.
• Cybersecurity: Real-time monitoring of network traffic to automatically identify code vulnerabilities and malicious packet patterns.
• Biotechnology: Processing and interpreting complex biological datasets for research or diagnostic purposes.
• Robotics & IoT: Converting environmental sensor data (LiDAR/visual) into a navigatable 3D mental map for physical traversal.

Example Prompts

1. "Scan the attached repository for any critical security vulnerabilities and generate a summary report of the findings."
2. "Perform a semantic deep-scan on these legal filings and identify any references to the merger negotiations from Q3."
3. "Analyze the biometric sensor logs from the lab and flag any readings that deviate from the expected physiological baseline."

Tips & Limitations

• Contextualization: Remember that Scan is optimized for sub-millisecond inference. Feed it high-quality data to ensure the internal knowledge graph cross-references are as accurate as possible.
• Compute Resources: Processing multi-modal data—particularly spatial mapping or genomic sequencing—is resource-intensive. Ensure your agent has adequate allocation.
• Limitations: Scan is a perceptual tool; it requires a higher-level planning agent to decide what to do with the extracted data. Avoid running deep-scans on encrypted data without the proper decryption keys provided in the context.