business-planner

• Outcome: Pass
• Latest audit: May 28, 2026, 11:32 AM
• Version:
  1.0.0

Overview

This is an instruction-only business planning helper that creates local documents and reuses workspace references, with no evidence of exfiltration or unsafe execution.

Install is reasonable if you want a local business-plan and pitch-deck generator. Before using it, confirm the workspace files it may read, review generated outputs for confidential business details and citation accuracy, and check the target filename before allowing saves or revisions.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (4)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill explicitly instructs saving generated business-plan output to a workspace path without stating that the agent should ask for user confirmation first. Automatic writes can create or overwrite files unexpectedly, which is risky in agent environments where workspace contents may be important and user intent may be ambiguous.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The infrastructure-diagram flow directs the agent to save output automatically to `projects/infra/[project]-infra.md` with no warning or approval step. Even though the content is benign, silent file creation/modification violates least surprise and could overwrite existing project artifacts or clutter the workspace.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The pitch-deck generation process includes an unconditional save step to `projects/pitch/[company]-pitchdeck.md` without any explicit consent or overwrite safeguards. In an automated agent setting, this can lead to unintended persistence of generated content and accidental replacement of existing files.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The version-comparison and feedback flows instruct the agent to read existing files and generate a new revised version (`v[N+1].html`) without a clear warning, approval step, or scope restriction. Unprompted reads can expose unrelated workspace data, and automatic writes can modify project state in ways the user did not explicitly authorize.

VirusTotal

66/66 vendors flagged this skill as clean.