Security audit

loop-constructor

Security checks for vulnerabilities and agentic risk

Overview

The skill is a design-only loop-planning helper that writes reviewable runbooks and does not execute the generated autonomous workflow.

Installers should understand that this skill produces runnable loop runbooks for later use. Review generated .loop files before running them, especially any restart route, and run resulting workflows only in an isolated workspace with version control checkpoints.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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 (1)

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: This section explicitly endorses autonomous restart behavior that discards a stage's work and re-derives it from the contract, but it does not pair that guidance with a clear safety warning or mandatory guardrails about preserving user work, capturing state, and obtaining approval for destructive actions. In the context of an agent-loop design skill, that omission can normalize designs that silently overwrite, revert, or discard meaningful work products when a loop decides patching has stalled.

Static analysis

No suspicious patterns detected.