Security audit

Singleshot Prompt Testing

Security checks across malware telemetry and agentic risk

Overview

This documentation-only skill coherently helps users install and use a prompt-testing CLI, but users should treat tested prompts, images, API keys, and generated reports as sensitive.

Install only if you intend to use the singleshot CLI. Verify the upstream Homebrew tap, Cargo crate, or source repository; use scoped provider API keys with spending limits; avoid testing secrets, customer data, confidential prompts, or sensitive images with third-party providers; and protect or delete generated reports because they may contain raw prompts and model outputs.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The README instructs users to send prompts to external providers and save reports locally, but it does not warn that prompts, attached images, and model outputs may contain sensitive data that will leave the local environment and be persisted to disk. In a prompt-testing skill, this omission is security-relevant because users may test production prompts, internal documents, or screenshots and inadvertently disclose secrets or regulated data.

Missing User Warnings

Low

Confidence: 84% confidence
Finding: The environment variable examples normalize placing live API keys in shell configuration without any guidance on secret handling, rotation, or avoiding accidental exposure through shell history, shared terminals, or committed dotfiles. While this is common documentation practice, in a security review it is still a real weakness because the skill encourages credential use without adjacent safeguards.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The release notes explicitly state that generated reports include full response content, but provide no warning that model outputs may contain sensitive prompts, secrets, personal data, or proprietary material. If users enable report generation during prompt testing, this can lead to unintended persistence of sensitive data in markdown files, logs, repositories, or shared artifacts.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill repeatedly instructs users to send prompts, configuration files, and even images to external model providers such as OpenAI, Anthropic, OpenRouter, and Ollama-compatible services, while also saving detailed reports locally, but it never warns about data exposure or local artifact creation. This can cause users to unknowingly transmit sensitive data to third parties and leave prompt contents, token usage, and outputs on disk where they may be retained, shared, or later disclosed.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The trigger phrases are broad, generic prompt-engineering terms such as 'test prompt', 'optimize prompt', and 'compare models'. In an agent ecosystem, these can cause the skill to activate for many unrelated user requests, leading to unintended invocation, context capture, or execution of prompt-testing functionality when the user did not explicitly request this specific skill.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.