Security audit

master-buddhaghosa

Security checks across malware telemetry and agentic risk

Overview

This is a coherent Chinese-language teaching skill for Theravāda/Buddhaghosa study, with some broad activation and language-rigidity concerns but no hidden sensitive access or harmful behavior.

Install this if you want a Chinese-oriented Buddhaghosa/Theravāda study assistant with strict citation behavior. Be aware that it may answer in a formal Chinese doctrinal style and may activate broadly for related Buddhist questions; use a qualified teacher rather than this skill for personal meditation instruction.

SkillSpector

By NVIDIA

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Intent-Code Divergence

Medium

Confidence: 81% confidence
Finding: Lines L169-L180 state that Buddhaghosa does not emphasize direct practical meditation guidance and that specific meditation guidance should uniformly be deferred to a qualified teacher. However, earlier sections give concrete practitioner-facing guidance such as matching meditation objects to temperament (L064-L072) and describing purification/progress stages (L076-L106), which goes beyond a purely conceptual framework and contradicts the claimed 'uniform response' policy.

Vague Triggers

Medium

Confidence: 97% confidence
Finding: This markdown file defines triggers, so SQP-1 applies. The phrase 'invoke whenever user's question touches Theravāda commentarial / Visuddhimagga / Abhidhamma exegesis, even without explicit request' is ambiguous and expansive, which could cause unintended activation for loosely related religious or academic questions.

Natural-Language Policy Violations

Medium

Confidence: 89% confidence
Finding: These lines prescribe a specific linguistic style and opening formulations for responses, effectively forcing one response mode regardless of the user's preference. Under SQP-3, fixed language or locale behavior without opt-in can be a policy violation unless choice or justification is clearly provided.

Natural-Language Policy Violations

Medium

Confidence: 90% confidence
Finding: This manifest presents the skill primarily in Chinese fields and sets tradition metadata in Chinese while also listing multiple languages, but it does not state any user-selectable language behavior or opt-in for locale. Under the policy rule, forcing a specific language/locale without user choice is a natural-language policy concern.

Natural-Language Policy Violations

Medium

Confidence: 79% confidence
Finding: The instruction at L179-L180 mandates a single fixed response in Chinese for a class of user queries. Under the policy, forcing a specific language without user opt-in is a natural-language locale violation unless the file clearly documents a justified region-specific constraint, which it does not.

Natural-Language Policy Violations

Medium

Confidence: 90% confidence
Finding: The instructions require a fixed Chinese reminder at the end of every reply and constrain first-turn address terms to Chinese forms such as “您 / 你 / 问者.” This is a natural-language locale policy constraint that applies regardless of user preference, and the file does not offer an opt-in or alternative language choice.

Natural-Language Policy Violations

Low

Confidence: 79% confidence
Finding: This markdown file uses Chinese as the sole instructional/documentation language across the entire document. Under the stated policy, forcing a specific language without giving the user a choice can be considered a natural-language policy violation unless the locale constraint is explicitly justified.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.