ClawHub
Back to skill

Security audit

Precc Token Saver

Security checks across malware telemetry and agentic risk

Overview

The skill is transparent about trying to save tokens, but it asks to rewrite every shell command and install/update external tooling in ways users should review carefully.

Install only if you trust PRECC enough to inspect and rewrite shell commands across your projects. Prefer a pinned or package-manager install over curl-to-bash, avoid broad history ingestion on sensitive work, and verify how to disable the hook, review rewritten commands, update safely, and clear learned local data.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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
Findings (3)

Vague Triggers

Medium
Confidence
92% confidence
Finding
The skill directs the agent to intercept and rewrite every shell command with no scoped exceptions, which gives an external helper broad influence over all command execution. In this context, the hook is LLM-enforced rather than OS-enforced, so the agent is being instructed to trust arbitrary command transformations globally, increasing the chance of unintended, unsafe, or attacker-influenced execution.

Missing User Warnings

High
Confidence
96% confidence
Finding
The skill explicitly tells the agent to execute a rewritten command from `precc-hook` instead of the user's original command, but it provides no user-facing disclosure or confirmation step. That creates a command integrity problem: a hidden rewrite layer can alter working directories, add wrappers, or substitute behavior in ways the user did not request and may not detect before execution.

Missing User Warnings

Medium
Confidence
99% confidence
Finding
The setup instructions fetch a remote script and pipe it directly into `bash`, which executes unreviewed network content immediately on the user's system. If the source, transport, repository, or upstream account is compromised, this becomes a straightforward remote code execution path.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.