The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external, untrusted data which creates a surface for indirect prompt injection.
Ingestion points: External DNS record values (TXT, CAA, DS, CNAME) are fetched from external servers in scripts/dns_security_checker.py.
Boundary markers: Absent. The logic does not encapsulate external data within protective delimiters to prevent instruction hijacking.
Capability inventory: The skill can execute system commands (dig, nslookup), perform network resolution (socket.getaddrinfo), and perform file write operations for reporting.
Sanitization: While regex is used for structural parsing, there is no content sanitization to filter out malicious LLM instructions.
[COMMAND_EXECUTION]: The skill utilizes system-level networking tools to gather audit data.
Evidence: scripts/dns_security_checker.py invokes subprocess.run to call dig and nslookup. These commands are executed with argument lists, mitigating shell injection risks, but they represent a significant capability that interacts with the host system.
[DATA_EXFILTRATION]: The skill performs network operations to reach external DNS servers.
Evidence: scripts/dns_security_checker.py performs lookups against arbitrary domains provided by the user via dig and socket.getaddrinfo. These operations are limited to DNS queries for auditing purposes and do not involve sensitive local file access.

infrastructure-compliance-auditor