Back to skill

Security audit

Multi Platform Publisher Agent

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent publishing assistant, but it asks agents to automate account login, public posting, scheduling, and third-party captcha solving without enough user control or data-handling limits.

Review this skill carefully before installing. Use it only with explicit user approval for each target platform, account, manuscript, upload, and publish or schedule action. Do not allow automatic captcha-solving unless you have verified it is permitted by the platform and acceptable for your privacy needs; prefer a human pause for verification. Avoid providing stored credentials or broad session access, and limit returned logs to the minimum publication status you need.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The activation scope is very broad, covering publishing, scheduling, formatting, captcha handling, and status reporting across multiple platforms without clear guardrails or trigger boundaries. In an agent setting, overbroad invocation increases the chance the skill is used for sensitive account actions, external posting, or policy-sensitive automation without adequate confirmation, scoping, or platform checks.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill describes transmitting book content, account data, cover assets, and possibly captcha-related data to external platforms and third-party APIs, but it does not clearly warn users about privacy, data-sharing, or retention implications. Because this skill operates across multiple publishing platforms and external services, the context increases the risk of inadvertent disclosure of sensitive account, manuscript, or metadata to parties the user may not expect.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill states it will not bypass or circumvent verification, yet its operational flow calls a third-party captcha-solving API to automatically complete slider challenges and continue execution. In context, this is effectively outsourced captcha bypass for automated account actions, which can violate platform anti-abuse controls, create compliance/legal risk, and expose account/session data to a third party.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill instructs logging and returning platform book IDs, links, review status, and publication timestamps, and the broader document also includes account configuration examples. In an agent workflow, these outputs can easily over-collect or echo sensitive identifiers, internal links, or account-associated metadata into logs, downstream tools, or chat responses, creating a practical data leakage channel.

Content

No source excerpt is available for this finding.