ClawHub
Back to skill

Security audit

topped

Security checks for vulnerabilities and agentic risk

Overview

The scanners report a skill for creating and posting short-form social media content, but flag broad activation wording, so ClawScan classifies it as suspicious with medium confidence; coverage was partial.

Before installing, check that you are comfortable with the skill’s activation wording and understand when it may act on your behalf. Ask for clear confirmation before consequential actions. The scan was partial: obfuscated instruction text could not be fully evaluated, several static checks were degraded, and some local references were unresolved.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

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 9)May include surrounding context.

md
# Topped

Topped makes short-form Posts (slideshows and videos) for an indie founder's Product, and posts them to the founder's TikTok, Instagram and YouTube accounts. You act as the founder: Topped has no approval step of its own for you, so the founder's word in this conversation is the approval.

The same tools run two ways. In a shell, use the `topped` CLI: each command runs one tool and prints one JSON document. In a chat app or any MCP client, call the tools by name on Topped's MCP server. Every recipe below gives both.

The activation description includes generic requests such as "fill my week" and questions about a "Library", "Calendar", or "Credits" without requiring a Topped or social-media context.

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The activation description includes ambiguous requests such as "fill my week" and references to a Library, Calendar, or Credits without explicitly requiring a Topped or social-media context. This could misroute unrelated requests into the skill, leading to unnecessary account queries or inappropriate workflow recommendations. Explicit consent requirements for publishing, deletion, and spending substantially limit the consequences.

Content

No source excerpt is available for this finding.