Security audit

Secretary Skill

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed goal-planning tool that stores local goal files and can prepare reminders, with privacy considerations users should understand before use.

Install only if you are comfortable with goal details being saved locally and potentially remembered across sessions. Use it with explicit save/schedule commands, avoid storing highly sensitive details unless needed, and periodically review or delete files under the configured goals directory and any reminders it creates.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (6)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill advertises persistent disk storage and memory retention as core behavior but does not clearly warn users at the start that personal goal details will be retained across sessions. Because goals may include financial, health, or career information, silent persistence materially increases privacy and data exposure risk.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The cron automation section instructs the agent to create recurring scheduled jobs, but it does not clearly warn the user that this will establish persistent system automation beyond the current conversation. Recurring jobs can generate unwanted reminders, consume resources, or be abused for persistence if created without explicit informed consent.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The method accepts a caller-supplied filepath and writes to it directly via Path(filepath) and _write(), with no restriction that the destination stay within the plans directory. If an untrusted agent or upstream input can control filepath, this enables arbitrary file overwrite within the permissions of the running process.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: add_alignment() silently modifies whatever filepath is provided, reading and rewriting the file contents without validating the path or requiring confirmation. In an agent context, this is risky because a prompt-controlled or tool-controlled filepath could cause unauthorized modification of unrelated files accessible to the process.

Ssd 3

Medium

Confidence: 95% confidence
Finding: The skill stores a cross-session memory pointer containing title, file path, status, target, and progress, then uses it later to recall user goals. Retaining and resurfacing personal planning data across sessions without explicit access controls or minimization increases the risk of privacy leakage, especially for sensitive financial or health-related goals.

Ssd 3

Medium

Confidence: 93% confidence
Finding: The skill includes commands to list all goal files, show full stored content, and summarize prior entries without any additional access-control or confirmation step. In a shared or misrouted session, this behavior could expose previously stored personal data broadly and make bulk disclosure easy.

Static analysis

No suspicious patterns detected.