Security audit

Sqlite Manager Free

Security checks for vulnerabilities and agentic risk

Overview

This is a local SQLite guidance skill with disclosed read/write/exec needs, but users should treat database-changing actions carefully.

Install this only if you want an agent to help manage local SQLite databases. Review commands before running writes, resets, imports, exports, VACUUM, or schema changes, and keep backups of important database files. The network troubleshooting line is not needed for normal local SQLite use.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (5)

Description-Behavior Mismatch

Low

Confidence: 84% confidence
Finding: The skill is documented as a fully local SQLite tool with no cloud/API-key dependency, yet it instructs agents to perform generic network diagnostics. This inconsistency can expand the tool's effective behavior beyond its stated scope and may cause an agent to execute unnecessary external commands in environments where network access should not be touched.

Context-Inappropriate Capability

Low

Confidence: 87% confidence
Finding: Including ping-based connectivity checks in a SQLite local-management skill grants an unjustified operational pathway unrelated to the stated purpose. Even if low severity, it encourages agents with exec capability to perform external-facing commands, increasing attack surface and creating opportunities for misuse or environmental probing.

Intent-Code Divergence

Low

Confidence: 82% confidence
Finding: The skill claims that data remains fully local and that no cloud service is involved, but later discusses network failures and ping-based remediation. Such contradictory instructions undermine operator expectations and can cause an agent to take actions outside the user's likely consent model for a local database tool.

Vague Triggers

Medium

Confidence: 77% confidence
Finding: The trigger condition 'use when database operations, SQL queries, or data storage management are needed' is overly broad for a skill with read, write, and exec capabilities. In agentic systems, ambiguous activation criteria can cause the skill to be invoked in unintended contexts, leading to unnecessary data access or modification on local databases.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill advertises modify, reset, import, export, and save operations but does not prominently warn that these actions can alter or overwrite local data. In a tool with write and exec permissions, lack of explicit mutation warnings increases the risk of accidental destructive operations by users or autonomous agents.

Static analysis

No suspicious patterns detected.