ClawHub
Back to skill

Security audit

data-infographic

Security checks for vulnerabilities and agentic risk

Overview

The skill can spend Canvora credits and submit your data without clearly requiring you to request or authorize image generation.

Before enabling this skill, restrict it to explicitly requested visuals and establish a credit budget or approval requirement. Only submit information approved for Canvora processing, and avoid confidential reports or URLs containing access tokens. The clean static scan does not resolve the authorization concern visible in the instructions.

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
  • 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 (2)

Activation includes merely having numbers, steps, or a roadmap, without requiring a request for a visual.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

The activation criteria include merely having numbers, steps, or a roadmap, which can route unrelated tasks into a paid visual-generation workflow without a request for visuals. This is more consequential because the skill instructs the agent to announce generation charges rather than ask permission, potentially causing unwanted spending and external data submission. Actual harm depends on whether the host agent independently enforces task scope and spending authorization.

Content

No source excerpt is available for this finding.

The workflow sends user-provided data and report URLs to Canvora without a privacy disclosure.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The commands submit user-provided text and report URLs to an external generation service without checking whether the material is sensitive or authorized for third-party processing. Reports and business figures can contain confidential information, and report URLs can themselves expose access tokens or access to private documents. Canvora processing is intrinsic to the advertised functionality, so this is a missing privacy boundary rather than evidence of covert or malicious exfiltration.

Content

No source excerpt is available for this finding.