Axiomata Vdv Core

• Outcome: Pass
• Latest audit: May 29, 2026 · 8:46 PM UTC
• Version: 1.0.1

Overview

This is a local text-analysis skill with optional user-directed file input and local JSON storage, with no evidence of hidden network access, credential use, destructive behavior, or automatic persistence.

Install only if you are comfortable running local Python helper scripts. Avoid using the storage options or agent memory with secrets, personal data, or private documents unless you intentionally want derived results retained in local JSON files.

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 (7)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 89% confidence
• Finding: The skill documentation advertises commands that persist results locally via `--store` and references a storage component, yet the skill metadata shown here declares no permissions. In an agent environment, undocumented file-write capability breaks least-privilege expectations and can enable unintended data retention or overwriting of local files if invoked automatically.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 95% confidence
• Finding: The declared purpose frames the skill as a general analysis methodology, but the documented behavior extends to local persistence, loading prior results, user-influenced file access via `@filepath`, and statistics/reporting functions. This mismatch is dangerous because operators or orchestrators may grant or invoke the skill under the assumption it is purely analytical, while it actually touches the filesystem and processes local data beyond the stated scope.

Description-Behavior Mismatch

Medium

• Confidence: 85% confidence
• Finding: The tool is presented as a general-purpose analysis engine, but it also persists analyzed content and metadata to a local JSON file without strong consent, sensitivity checks, or retention controls. If users analyze secrets, internal documents, or personal data, the script can create unintended local data stores that expose sensitive material to other local users, backups, or later exfiltration.

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: The `@filepath` feature allows the caller to read arbitrary local files and feed their contents into the analyzer, which is broader than the stated purpose of a truth-analysis protocol. In an agent or automation context, this expands the skill into a local file access primitive that could be abused to inspect sensitive files and then print or store derived data.

Description-Behavior Mismatch

Medium

• Confidence: 88% confidence
• Finding: The script persistently stores analysis content to a local JSON file, which expands the skill from transient analysis into data retention. In an agent setting, analysis content can contain sensitive prompts, secrets, user data, or derived internal state, so undocumented persistence creates confidentiality and privacy risk even if the implementation is simple.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The user-controlled --path parameter allows arbitrary local file read/write within the permissions of the executing process. An attacker or untrusted caller could overwrite application files, write into sensitive locations, or read arbitrary JSON files through the load path, which is especially dangerous because this storage behavior is unrelated to the stated truth-analysis purpose.

Missing User Warnings

Low

• Confidence: 82% confidence
• Finding: The code writes user-supplied content to disk without any explicit warning, confirmation, or safeguard around persistence and overwrite behavior. In an agent workflow, users may assume analysis is ephemeral, so silent retention can leak sensitive content and create unexpected forensic traces or data handling issues.

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.