Prd Skill

Security checks across malware telemetry and agentic risk

Overview

The skill is a plain markdown PRD generator, but it also instructs the agent to save outputs and automatically hand them to downstream development skills without an approval checkpoint.

Install this only if you want PRD requests to feed an automated development pipeline. For safer use, instruct the agent to return the PRD first and require explicit approval before saving files, sending session messages, or invoking dev-skill or qa-skill.

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 (2)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The skill description uses very broad activation language ('Use when a user provides an app idea or feature request') without clear boundaries, exclusions, or confirmation requirements. In an agentic system, this can cause over-invocation on loosely related prompts, unnecessarily exposing user content to the skill and increasing the chance of unintended pipeline execution.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill documents automatic side effects: saving files to `prd-output/` and triggering `dev-skill` plus session messaging, but does not indicate any user-visible warning, consent, or approval step. This is dangerous because simply using the skill may persist user data and launch additional automated actions, creating risks of unauthorized data handling, unexpected execution chains, and privilege expansion across connected skills.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal