Security audit

Vue

Security checks for vulnerabilities and agentic risk

Overview

This is a Vue guidance skill with scoped, disclosed preference storage and no executable payload or hidden high-impact behavior.

Before installing, be aware that the skill keeps Vue-related preferences and memory under ~/Clawic/data/vue/. Review or delete those files if you do not want persistent project or style preferences carried between sessions.

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

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

Findings (7)

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 80% confidence
Finding: Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Content: Scanner excerpt · errors.md (reported line 3)May include surrounding context.
md
# Errors — Boundaries, the App Handler, and Failure States Vue ships no default error UI. An uncaught throw in setup or render unmounts that subtree and leaves a blank area plus one console line — no warning, no fallback. Every layer below is opt-in, and a production app needs all three: a global handler, boundaries around independently-failing regions, and a designed failure state per async widget. ## What Catches What

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 80% confidence
Finding: Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Content: Scanner excerpt · reactivity.md (reported line 31)May include surrounding context.
md
# Errors — Boundaries, the App Handler, and Failure States Vue ships no default error UI. An uncaught throw in setup or render unmounts that subtree and leaves a blank area plus one console line — no warning, no fallback. Every layer below is opt-in, and a production app needs all three: a global handler, boundaries around independently-failing regions, and a designed failure state per async widget. ## What Catches What

Hidden Instructions

High

Category: Prompt Injection
Confidence: 70% confidence
Finding: Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
Content: Scanner excerpt · memory-template.md (reported line 13)May include surrounding context.
md
last: YYYY-MM-DD ## Context   ## Pain Points

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 75% confidence
Finding: Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
Content: Scanner excerpt · SKILL.md (reported line 162)May include surrounding context.
md
- **conventions** — `ref` vs `reactive` house style, component and composable naming, SFC block order, feature-folder vs type-folder layout - **tooling** — build tool, package manager, UI kit, validation and i18n libraries, whether VueUse is welcome or the project prefers hand-rolled composables - **safety posture** — `v-html` policy, whether to run codemods without confirmation, how loudly to flag missing keys and unbounded watchers - **output format** — whether components ship with a test, how much explanation accompanies code, comment density - **platform** — browser support target, CSP strictness (which decides runtime-only vs compiler build), SSR host

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill explicitly instructs the agent to read files from the user's home directory (~/Clawic/data/vue/config.yaml and ~/Clawic/data/vue/memory.md) automatically on first use, without any disclosure or consent step. Even though the files are framed as preferences, they may contain sensitive personal or project context, and silent access increases the risk of privacy violations and unintended data exposure across tasks.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill directs the agent to persist user-provided preferences and behavioral details into local config and memory files without notifying the user that their statements may be stored beyond the current session. This creates a privacy and consent problem, and can also propagate sensitive or security-relevant user data into long-lived storage that may later be read by other tasks or agents.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 83% confidence
Finding: The markdown instructs the user to move data from old locations into ~/Clawic/data/vue/, which is a file-system modifying operation affecting user data. The surrounding description does not explicitly warn about the action's impact, such as overwriting or relocating existing files, so users may not realize it changes local data.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.