Anthropic Cost Optimizer

Security checks across malware telemetry and agentic risk

Overview

This skill is not malicious, but it needs review because broad triggers can make it read and later rewrite live OpenClaw configuration files that may affect credentials, costs, and agent behavior.

Install only if you specifically want an OpenClaw cost audit. Before allowing it to read from disk, confirm the exact config path; before approving any write, review the diff, make a backup, and ensure the output stays valid for the original file type. Redact secrets if you paste config content.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger conditions are broad enough to activate on common billing, pricing, or config-help conversations, causing the skill to read local configuration files and prepare write-back changes in situations where the user may not have explicitly requested file access or modification. Because this skill searches several default paths and is designed to rewrite the live config, over-triggering increases the chance of unintended file access, confusing autonomous edits, and unnecessary exposure of sensitive configuration contents.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal