Security audit

Publisher

Security checks across malware telemetry and agentic risk

Overview

This is a publishing helper that does high-impact public publishing actions, but the reviewed artifacts disclose that workflow and keep it aligned with the skill’s purpose.

Install only if you want a tool that can edit SKILL.md, initialize git, commit local files, create a public GitHub repository, push code, and publish to ClawdHub. Before approving publication, inspect the skill directory, remove secrets or private files, and make sure .gitignore excludes anything that should not be published.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 94% confidence
Finding: The skill is presented primarily as a documentation/publishing helper, but it also performs sensitive side effects including modifying local files, creating repositories, pushing code to GitHub, and publishing externally to ClawdHub. That mismatch is dangerous because users may invoke it expecting content generation while unintentionally authorizing repository creation or exfiltration of code/metadata to third-party services.

Description-Behavior Mismatch

Medium

Confidence: 96% confidence
Finding: The skill’s stated purpose emphasizes documentation and visibility, but the script goes on to create a public GitHub repository and publish content to an external registry. That broader capability matters because it can expose the entire local skill directory and metadata to third parties, which is riskier than the description suggests and can lead to unintended public disclosure.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README promotes one-command publishing to GitHub and ClawdHub and repeatedly emphasizes automation, but it does not prominently warn that local skill contents will be pushed to external services. Although there are sample prompts for approval later, the top-level messaging can cause users to run the tool without fully understanding the data exposure and repository creation implications.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill describes actions that alter repository state and transmit content to external services, but it does not foreground the risks of modifying SKILL.md/README, committing all files, pushing to GitHub, or publishing to ClawdHub. In practice, this can cause accidental disclosure of private code, secrets, or unfinished work if a user treats it as a low-risk documentation utility.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The publish step uploads the local skill directory to an external service, but the skill description does not clearly communicate that data transfer and publication occur. In context, this is dangerous because users may run the tool expecting local documentation help, not realizing their repository contents may be transmitted or made public.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.