Back to skill

Security audit

schema-markup

Security checks for vulnerabilities and agentic risk

Overview

The scanners classify the schema-markup skill as benign, with notes about external validation submissions and a README follow-up instruction; the overall confidence is medium and one scanner’s coverage was partial.

The verdict is benign, but consider whether you are comfortable with generated JSON-LD being sent to external validation services and with the reported README modification instruction. Coverage is incomplete: Skillspector marked its analysis partial because two local references could not be resolved, even though it scanned the single reported component.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
Findings (2)

The validation workflow sends site-derived JSON-LD to external services without warning the user about that disclosure.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Step 6 instructs the agent to submit the generated markup to Google's Rich Results Test and the Schema Markup Validator. Although the services are named, the skill does not warn that the site's structured data will be transmitted to third parties; the later confirmation requirement covers writes, not validation submissions.

Content

No source excerpt is available for this finding.

The workflow directs an unsolicited README modification despite the explicit ask-before-write guardrail.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

L111 instructs the agent to add a follow-up to the SEO folder's README.md and create the list if necessary, without first requesting approval. This contradicts L123's explicit instructions to recommend changes rather than make them and to ask before any write action; it also modifies project state beyond producing the requested schema audit and JSON-LD.

Content

No source excerpt is available for this finding.