Security audit

Works: you projects organized.

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local project-notes organizer that clearly creates, updates, links, and moves project files, with no evidence of hidden exfiltration or destructive behavior.

Install this only if you want your agent to maintain durable local project records. Review the GitHub source before using raw install or update commands, back up mind/works/, and configure your agent to ask before creating, moving, or cross-linking files if you prefer tighter control.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Context-Inappropriate Capability

Low

Confidence: 95% confidence
Finding: The skill includes instructions to fetch content from GitHub and replace the local skill file, which introduces network access and supply-chain risk outside the core local project-management function. Even if intended for maintenance, a compromised remote source, MITM, or repository change could alter agent behavior without adequate review.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The README describes creating per-project folders, maintaining logs, and moving completed works to a closed directory, but it does not warn that the skill modifies the user's local filesystem and project records. In an agent ecosystem, this omission can cause users or integrators to underestimate the write impact, increasing the chance of unintended file creation, reorganization, or data changes from casual prompts.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The activation criteria are broad and imperative, telling the agent not to ask permission and to open a work whenever conversation suggests multi-session activity. This can cause unintended file creation/modification during ordinary conversation, especially when the user is brainstorming rather than requesting persistence.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The skill instructs the agent to act automatically whenever the human mentions progress, including appending logs, updating next steps, routing artifacts, and editing related files. Without strong trigger constraints or confirmation, this can lead to unauthorized or inaccurate workspace changes from casual mentions or ambiguous statements.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill description and early instructions emphasize behavior but do not clearly warn that it will create directories, write work.md files, update links, and move folders in the workspace. Users may invoke it expecting organizational help without realizing it performs persistent file operations automatically.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The update instructions tell the agent to fetch remote content and replace the local SKILL.md, but omit any warning about network access, trust boundaries, or the danger of replacing executable instructions with unreviewed remote data. This creates both transparency and supply-chain problems because the user may not realize the skill can self-alter from an external source.

Static analysis

No suspicious patterns detected.