The Agent Skills Directory

[METADATA_POISONING]: There is a discrepancy between the author listed in the SKILL.md frontmatter ('mahipal') and the copyright holder specified in the LICENSE file ('mukul975'). Such inconsistencies in metadata can be deceptive regarding the origin and ownership of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from phishing emails, which is a known vector for indirect prompt injection.
Ingestion points: The scripts/agent.py script reads external email files (.eml) via the --email-file parameter.
Boundary markers: No delimiters or explicit instructions are used to prevent the agent from following instructions potentially embedded within the phishing email content.
Capability inventory: The skill has the capability to perform network requests to an external SOAR API and write results to the local filesystem.
Sanitization: The script extracts URLs, IPs, and headers using regex and basic string slicing, but it does not perform robust sanitization or escaping of the content before it is processed or sent to the SOAR platform.
[CREDENTIALS_UNSAFE]: The scripts/agent.py script includes a --no-verify flag that disables SSL/TLS certificate verification for all REST API requests. When enabled, this allows the SOAR API token (transmitted in the ph-auth-token header) to be sent over insecure connections, making it vulnerable to capture via man-in-the-middle (MITM) attacks.
[COMMAND_EXECUTION]: The script executes network commands using the Python requests library to interact with a user-provided SOAR instance URL.
[EXTERNAL_DOWNLOADS]: The skill makes outbound network requests to communicate with the Splunk SOAR REST API.

implementing-soar-playbook-for-phishing