The Agent Skills Directory

[SAFE]: The skill follows a local-first architecture. It reads and writes only to specific local directories ('hunter/') and does not include any network-related code (such as requests, urllib, or curl calls), preventing data exfiltration of sensitive profile information.
[SAFE]: All tool logic is contained in 'answers.py', which uses standard library Python modules for string matching and JSON handling. There is no evidence of dynamic code execution, subprocess spawning with untrusted input, or remote code downloads.
[SAFE]: The skill instructions emphasize human-in-the-loop safety, specifically directing the agent to stop before submitting forms and prohibiting the handling of passwords or CAPTCHAs.

hunter-apply