ClawHub
Back to skill

Security audit

Art Critique

Security checks for vulnerabilities and agentic risk

Overview

This art critique skill creates local critique notes and copies submitted artwork into session folders, which is disclosed and fits its purpose.

Install this only if you are comfortable with your submitted artwork and critique history being stored locally in an art-critique-sessions folder. In shared or sensitive workspaces, ask the agent where it will write notes before starting, and delete session folders when you no longer need them.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The trigger phrases are broad and include common conversational language like "critique," "critic," and "talk about your feeling about my work," which can cause the skill to activate in contexts beyond explicit user intent. Because the skill has Read, Write, Edit, and Bash permissions, accidental activation increases the chance of unnecessary file creation or modification without the user expecting tool use.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
85% 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 · SKILL.md (reported line 173)May include surrounding context.

md
any **parked** thread to resume. Only re-read the FULL notes file every ~3 rounds, or when
   you sense you've drifted from the agenda — this keeps token cost down.
2. **DISCUSS.** Talk about that ONE aspect: share your genuine feeling/observation and
   **ask the artist for their view**. Go as deep as possible before moving on. Do not lecture
   or correct (see the Principle section).
3. **ASSOCIATE** (only when something genuinely fires). Follow `references/association-engine.md`:
   check whether the link is in the work or you'd be projecting it; pull at most **0–2** cards

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README advertises activation phrases like '帮我 critique 一下', 'review 我的作品', and '你对这件作品有什么感受', which are broad conversational requests rather than tightly scoped commands. In an agent environment that auto-selects skills by natural language matching, these generic phrases can cause unintended invocation and expose user content or trigger file/session creation when the user did not explicitly intend to use this skill.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The English examples such as 'critique this', 'help me review my artwork', and 'what do you feel about this piece' are common, everyday prompts that overlap with normal conversation. In systems using heuristic routing, this increases the chance of accidental skill activation, which can lead to unintended processing of user media and creation of persistent session artifacts.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill instructs the agent to create session folders, copy user files, and continuously update notes on disk, but it does not require explicit user consent or a clear user-facing notice before persistent storage occurs. This creates a privacy and integrity risk because sensitive creative works, transcripts, and critique history may be written locally by default, potentially without the user's awareness.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.