The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python helper script in §10 (spf_lookup_count.py) that utilizes the subprocess module to run the dig command. This utility is intended for DNS lookups and uses argument lists to call the process, which is a safer alternative to shell execution, though it still performs local command execution as part of its core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill contains logic for processing external DNS records, which are untrusted inputs. This creates a potential surface for indirect injection via crafted records intended to influence the agent's analysis.
Ingestion points: DNS TXT records for SPF and DMARC fetched from arbitrary domains as described in SKILL.md and the helper script.
Boundary markers: Reasoning instructions are provided for the agent to parse records, but there are no specific markers or delimiters defined to isolate untrusted DNS record content from the agent's instructions.
Capability inventory: The skill allows for reasoning about email security, manual installation of a Python script to the file system, and execution of the dig utility for network DNS queries.
Sanitization: The included Python script performs record parsing using basic regular expressions and string splitting.

email-domain-security