ClawHub

Writer

Security checks across malware telemetry and agentic risk

Overview

This Writer integration is not clearly malicious, but it gives the agent broad authenticated API authority through Membrane without tight scoping or strong user confirmation guidance.

Review this skill before installing if your Writer account contains sensitive company, customer, or unpublished content. Prefer using prebuilt Membrane actions, require explicit approval before any raw proxy request, and avoid DELETE/PATCH/POST calls unless the target endpoint and payload are clear.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The manifest frames the skill as interacting with Writer data, but the body expands scope to generic workflow automation through Membrane, including dynamic connection creation and direct API proxying. That mismatch can cause an orchestrating agent or user to invoke the skill under narrower trust assumptions than the skill actually requires, increasing the chance of unintended external actions or data movement.

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The raw proxy interface permits arbitrary HTTP methods and endpoints via an authenticated connection, which materially broadens capability beyond a constrained Writer integration. In an agent setting, this can enable unreviewed reads, writes, deletions, or sensitive data transmission to external API paths that were not intended by the user or by higher-level policy.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The invocation cue is broad enough that an agent may select this skill for loosely related requests without clear constraints on when it is appropriate. Over-broad routing increases the risk that the skill's powerful network and proxy capabilities are used in contexts where the user expected a narrower Writer-only action.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The proxy section explains how to send arbitrary requests but omits a clear warning that data and prompts may be transmitted over the network to external services using authenticated credentials. In an agent workflow, that missing disclosure increases the chance of accidental exfiltration of sensitive content or execution of external side effects without informed user awareness.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal