Agent Advisor
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This skill appears to be a read-only local advisor, but it does inspect recent OpenClaw conversation history and local OpenClaw configuration data.
This looks safe to install if you want a local model/security advisor. Be aware that the automatic recommendation mode reads recent OpenClaw session messages, and the security mode reads your local OpenClaw configuration; use the task-description mode instead if your recent history is sensitive.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The skill may read up to recent local user messages to recommend a model, though the shown code does not transmit or store them elsewhere.
The script reads recent local OpenClaw session files and extracts user message text to infer task patterns. This is disclosed and bounded, but conversation history can contain private information.
function loadHistoryMessages(maxSessions = 5, maxMsgsPerSession = 50) ... event.message?.role === 'user'
Use the history-based mode only if you are comfortable with the skill inspecting recent OpenClaw conversations; prefer task-description mode for sensitive histories.
The skill can inspect local OpenClaw configuration details, but the provided code does not show credential disclosure, mutation, or network sharing.
The script reads the local OpenClaw configuration file to calculate a security score. Such files may include authentication settings, but the shown output only uses configuration fields such as auth mode, bind address, gateway mode, and deny-command counts.
const CONFIG_PATH = join(HOME, '.openclaw', 'openclaw.json'); ... return JSON.parse(readFileSync(CONFIG_PATH, 'utf8'));
Review the script before use if your openclaw.json contains secrets, and avoid sharing generated reports if they reveal sensitive configuration details.