Back to skill

Security audit

Styling-Studio

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this styling-studio skill suspicious because scanners identified concerns about retaining conversation-derived content and broad activation triggers, while reporting no AIG findings.

Review these privacy and activation concerns before installing. In particular, consider whether you are comfortable with conversation-derived information being reused in a persistent library and with broad triggers potentially selecting the skill for unrelated requests. The scan reports complete component coverage, but a clean AIG result does not remove the concerns identified by Skillspector.

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
Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (3)

Conversation-derived content is reused in the persistent asset library without a privacy warning.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The skill directs search and conversation outputs back into a persistent, reusable asset library without specifying consent, de-identification, or separation of private profiles from reusable assets. Because its diagnostic workflow collects personal characteristics and potentially health-related skin information, generated materials could retain sensitive details and expose them through later library reuse; no external exfiltration destination is shown.

Content

No source excerpt is available for this finding.

The trigger list includes terms that are not specific to personal styling.

Low
Category
Trigger Abuse
Confidence
91% confidence
Finding

The trigger list includes “meta-skill-system,” which is unrelated to personal styling and could cause this skill to be selected for framework-related requests. This creates a limited routing-confusion risk: unrelated conversations could receive styling-specific instructions and workflows, although the file does not demonstrate deliberate routing hijacking or guaranteed automatic invocation.

Content

No source excerpt is available for this finding.

L0005 includes the cross-domain triggers “商家方案” (business plans) and “meta-skill-system” without requiring a styling-related request.

Low
Category
Trigger Abuse
Confidence
91% confidence
Finding

The description lists the unrelated trigger “meta-skill-system” and the broad trigger “商家方案” without requiring styling context, creating a risk of unintended skill selection for unrelated requests. Business-plan support is legitimate within this skill’s salon workflow, but these unqualified triggers can divert other business or agent-framework tasks into its styling pipeline. This supports a low-impact routing concern, not evidence of malicious execution or data theft.

Content

No source excerpt is available for this finding.