Security audit

Finanças Donna

Security checks for vulnerabilities and agentic risk

Overview

This personal finance skill is not malicious, but it should be reviewed because it can read and update sensitive finance files from very broad trigger words and contains one person's private financial context.

Install only if this is meant for Evandro's own finance workspace or after replacing the personal rules with your own. Use it with backed-up finance files, narrow the activation triggers if possible, and require explicit confirmation before any financial file is read or changed.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The skill declares automatic activation on very broad finance-related terms such as common verbs and category words, which can cause the agent to load and apply financial-file reading or update logic during ordinary conversation. In this context, overbroad triggering increases the chance of unintended access to sensitive financial data or accidental modification flows without sufficiently explicit user intent.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: Using generic file-processing triggers like 'fatura', 'extrato', 'cartão' and 'banco' can invoke the skill in many unrelated contexts, especially because the skill then instructs the agent to read financial files and process documents. That creates a realistic risk of unintended handling of banking/card materials or scope creep into sensitive documents without clear, explicit authorization.

Natural-Language Policy Violations

Low

Confidence: 81% confidence
Finding: The skill is explicitly built around a named individual and embeds highly specific personal relationships and financial rules, which can cause the agent to apply one person's private assumptions to whoever triggers the skill. In a shared or reused environment, this creates privacy and integrity risks by exposing personal context and by misclassifying another user's financial data based on hard-coded identity-specific logic.

Static analysis

No suspicious patterns detected.