The Agent Skills Directory

[OBFUSCATION]: The skill encourages the use of visually deceptive lookalike domains to bypass user scrutiny during phishing simulations.
Evidence: The file SKILL.md suggests using sender domains like rnicrosoft.com (spoofing microsoft.com) and cornpany.com (spoofing company.com).
Security Alert: Automated scanners have identified rnicrosoft.com as a domain associated with active phishing campaigns.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting untrusted external data, creating a potential attack surface if the data influences agent behavior.
Ingestion points: The skill describes using OSINT output from recon-agent as input for targeting and pretext generation in SKILL.md.
Boundary markers: No specific delimiters or security instructions are provided to isolate untrusted external OSINT data from the phishing templates.
Capability inventory: The skill provides methodology and infrastructure guidance for credential harvesting and payload delivery but does not include direct execution scripts.
Sanitization: The skill lacks instructions for sanitizing or validating external input data before it is incorporated into campaign pretexts.

phishing-sim