ClawHub
Back to skill

Security audit

AFOL Rebrickable

Security checks across malware telemetry and agentic risk

Overview

This Rebrickable skill uses expected API credentials and guarded collection tools, with risky reference-only endpoints not exposed by its CLI instructions.

Install only if you are comfortable providing a Rebrickable API key and, for private collection features, a user token. Keep the API base URL at the default, use dry-run before writes, and only approve mutations after the agent states the exact list, set or part, quantity, and create/update/delete action.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
91% confidence
Finding
The OpenAPI spec exposes a credential-minting endpoint (/users/_token/) that accepts username and password and returns a user token, expanding the skill from catalog lookup and guarded collection writes into full authentication handling. In an agent setting, this materially increases risk because the skill could solicit or process raw user credentials and then access broad account-management endpoints with the minted token, which is far beyond the narrowly described scope.

Description-Behavior Mismatch

High
Confidence
95% confidence
Finding
The /users/{user_token}/sets/sync/ endpoint performs full collection synchronization and explicitly states it will remove sets not present in the submitted list, making it a destructive bulk-write capability. That exceeds 'guarded collection writes' because a single mistaken or manipulated agent action could wipe or significantly alter a user's collection at scale rather than making a narrow, confirmable change.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
Although the description mentions replacement behavior, the operation summary and metadata do not prominently communicate that this endpoint is destructive and can remove existing user sets. In agent-mediated use, weak warning signals increase the chance that a model or integrator treats it like an ordinary update call, causing accidental mass deletion of collection data.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.