Skillv1.0.1

ClawScan security

System Monitor · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 7, 2026, 11:05 PM

Verdict: Benign
Confidence: high
Model: gpt-5-mini
Summary: This skill is internally consistent: it runs local system monitoring commands and a small bundled shell script, and it does not request credentials or external network access.
Guidance: This skill appears to do what it says: run local monitoring commands and return system status. Before installing, (1) verify the host has the CLI utilities you need (or adjust the SKILL.md to match your platform), (2) inspect the included scripts yourself (they are short and readable), and (3) be aware that running the skill reveals local process lists and network port/connection info — avoid running it on multi-tenant/shared systems where that information is sensitive. If you want stricter control, limit the agent's ability to invoke the skill autonomously or run it manually as needed.

Review Dimensions

Purpose & Capability: noteThe skill's name/description match its implementation: it runs local CPU, memory, disk, network, and process commands and includes a short script. Slight mismatch: the registry 'required binaries' list includes core tools (top, df, vm_stat, free, lsof, uptime) but the SKILL.md also references additional utilities (mpstat, hostinfo, ifconfig, nettop, iostat, ip, ss, pstree, du, swapon, awk, sort, etc.). These are reasonable for a system-monitor tool but are not enumerated in the metadata.
Instruction Scope: noteInstructions direct the agent to execute local CLI diagnostics only (top, df, vm_stat, free, lsof, uptime, ps, ifconfig/ss/ip, etc.). There are no commands that send data to external endpoints and no access to unrelated filesystem paths or environment secrets. Note: the commands collect local process and network information (process lists, listening ports, connections), which can be sensitive on shared systems—this is consistent with the stated purpose but worth awareness.
Install Mechanism: okNo install spec (instruction-only) and a single small shell script are provided. Nothing is downloaded or extracted from external URLs. Risk from installation is low.
Credentials: okThe skill requires no environment variables or credentials. It does not request unrelated secrets or config paths.
Persistence & Privilege: okThe skill is not marked always:true, does not request persistent system changes, and does not modify other skills' configuration. The agent may invoke it autonomously by default (ordinary behavior) but it does not gain elevated persistent privileges.