Security audit

Ghost Catalog - Semantic File Organization

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local file-cataloging helper whose reads and writes are mostly disclosed and aligned with its purpose.

Install only if you want the agent to maintain catalog metadata in your workspace. Run scans first, review any proposed tagging carefully, and avoid applying headers to files whose syntax or tooling may break when content is prepended, such as strict config files, secrets, generated files, or schema-validated JSON.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

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)

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill describes tagging behavior without prominently warning that it prepends headers into existing files, which is a direct content modification. Users may invoke tagging under the assumption it is metadata-only, leading to unintended source changes, file corruption in edge cases, or churn across many files if applied in bulk.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The auto-invocation guidance is overly broad and includes common phrases like asking what files are in a project, which can cause the skill to trigger in contexts where the user did not intend catalog operations. Because this skill can subsequently scan the workspace and offers write-capable actions, over-triggering increases the risk of unnecessary file enumeration and escalation into unintended modifications.

Missing User Warnings

Low

Confidence: 89% confidence
Finding: The `report` command writes a markdown file into the workspace, but this side effect is not clearly surfaced where users would expect operational warnings. Hidden write behavior is less severe than source-file rewriting, but it can still create unexpected files, overwrite prior reports, or affect automation that watches the docs directory.

Static analysis

No suspicious patterns detected.