The Agent Skills Directory

[DATA_EXFILTRATION]: The scripts/agent.py script reads external .eml files and transmits extracted data including headers, sender information, and email body contents to a remote SOAR instance defined by the user. While this is the intended function, it represents a data flow from local files to a network endpoint.
[COMMAND_EXECUTION]: The script implements a --no-verify command-line argument that disables SSL/TLS certificate verification for all REST API requests to the Splunk SOAR instance. Using this flag in production environments could allow an attacker to perform Man-in-the-Middle (MitM) attacks to intercept sensitive API tokens and incident data.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from external sources.
Ingestion points: scripts/agent.py parses email headers and body content from files provided via the --email-file parameter.
Boundary markers: No boundary markers or protective instructions are used to separate untrusted email content from agent instructions.
Capability inventory: The script performs network POST operations to a remote API and reads local files.
Sanitization: Indicators are extracted using regular expressions and truncated for length, but the content is not sanitized for malicious natural language instructions that might influence an AI agent processing the output.

implementing-soar-playbook-for-phishing