schema-markup-audit
Warn
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The provided scripts in 'references/checks.md' use the '$URL' variable within shell commands. If an agent populates this variable using unsanitized user input, it could allow an attacker to execute arbitrary commands by appending shell metacharacters to the URL string.
- [DATA_EXFILTRATION]: The auditing workflow relies on 'curl' to access user-supplied URLs. This capability enables Server-Side Request Forgery (SSRF), which can be used to probe internal networks or harvest credentials from cloud metadata services (e.g., AWS, GCP, or Azure metadata endpoints).
- [PROMPT_INJECTION]: This skill's primary function is to audit data from external sources, presenting a significant surface for indirect prompt injection. An attacker could place malicious instructions inside a website's schema.org JSON-LD to manipulate the agent's behavior when it parses the content. Ingestion points: Data extracted via 'curl' in 'references/checks.md'. Boundary markers: None defined to isolate external content. Capability inventory: Network access via 'curl', shell processing with 'grep' and 'sed', and execution of inline Python scripts. Sanitization: No evidence of validation, escaping, or instruction-filtering for the ingested data.
Audit Metadata