Nonprofit Grant Outline

• Outcome: Review
• Latest audit: May 29, 2026 · 9:26 PM UTC
• Version: 1.0.0

Overview

This skill appears to be advertised as a grant-proposal helper but includes broad local auditing and scanning behavior that users may not expect.

Review this carefully before installing. Only use it in a disposable or non-sensitive workspace unless the publisher updates the description and metadata to clearly disclose the auditing modes, narrow the accepted paths and file types, and require explicit user approval before scanning directories or writing reports.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 90% confidence
• Finding: The skill advertises and enables code-capable behavior via `python3` execution and references to local input/output files, but does not declare corresponding permissions or constraints. This creates a trust and review gap: operators may assume the skill is purely text-generation while it can read local resources, invoke a script, and write output files.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 97% confidence
• Finding: There is a substantial mismatch between the stated purpose (generating nonprofit grant proposal outlines) and the reported behaviors (directory scanning, CSV/TSV analysis, risky-pattern scanning, skill structure auditing, and writing Markdown/JSON results). Hidden or unrelated capabilities are dangerous because they expand the attack surface and can be abused to inspect local data or perform unexpected filesystem actions under a benign-looking description.

Description-Behavior Mismatch

High

• Confidence: 97% confidence
• Finding: The script’s primary behavior is a generic local auditing/scanning tool selected by spec-driven modes, including directory, CSV, pattern, and skill audits, rather than a narrowly scoped nonprofit grant-outline generator. In the context of a grant-writing skill, this capability mismatch is dangerous because it enables broad inspection of local files and repositories the user points it at, creating unnecessary access to sensitive content and expanding the blast radius far beyond the stated purpose.

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The code embeds regex-based scanning for secrets, private URLs, and dangerous shell patterns, which is unrelated to drafting grant proposals and gives the skill a covert security-audit flavor. Even though it only reports matches, this can still expose snippets from sensitive files and encourages users to run the skill over repositories or documents containing credentials or internal paths.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The repository/skill structure audit logic checks for expected files, parses SKILL.md frontmatter, and reports repository metadata, which is functionally unrelated to nonprofit proposal outlining. In this context, it unnecessarily broadens the skill into a repo-inspection tool and can reveal internal project structure and metadata from arbitrary local directories.

VirusTotal

65/65 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.