Security audit

create-patent-based-rd-briefing-rd

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent local patent-workbook reporting tool with expected file processing behavior and no evidence of hidden network access, credential use, or destructive actions.

Install only into a directory you choose, review requirements.txt before using --install-dependencies, and use only trusted topic configuration files. Treat patent workbooks as confidential local data, confirm rights for embedded figures and links, and rely on the generated report only after human review and any required patent-professional review.

SkillSpector

By NVIDIA

Vulnerability Patterns

Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (4)

Description-Behavior Mismatch

High

Confidence: 96% confidence
Finding: The file is a hard-coded coffee-machine example configuration embedded in a skill whose declared purpose is generating patent-based R&D briefings from authorized user-supplied workbooks. This domain mismatch can cause the agent to apply the wrong keyword filters, categories, and entity aliases, leading to systematic mis-screening of patents, omission of relevant records, and inclusion of unrelated material while presenting the output as a reproducible professional briefing.

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The script loads topic configuration files as Python modules and executes them via spec.loader.exec_module(), which gives any code in those files full execution privileges at report-generation time. In a skill that is expected to turn an authorized workbook into HTML, this expands the trust boundary from data-only inputs to arbitrary executable code, enabling local code execution, file access, or process launching if a config file is modified or replaced.

Unpinned Dependencies

Low

Category: Supply Chain
Content: pandas>=2.0.0 openpyxl>=3.1.0
Confidence: 91% confidence
Finding: pandas>=2.0.0

Unpinned Dependencies

Low

Category: Supply Chain
Content: pandas>=2.0.0 openpyxl>=3.1.0
Confidence: 91% confidence
Finding: openpyxl>=3.1.0

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.dynamic_code_execution

Dynamic code execution detected.

Critical

Code: suspicious.dynamic_code_execution
Location: scripts/generate_report.py:137

Dynamic code execution detected.

Critical

Code: suspicious.dynamic_code_execution
Location: scripts/tag_relevant.py:102