The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the rclone utility to transfer files from the local filesystem (/tmp/) to a remote Google Drive destination. While this is part of the intended workflow for saving outreach drafts, it involves executing shell commands with parameters derived from agent-generated content.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it retrieves data from external job boards and company websites via web_search.
Ingestion points: Data enters the agent context through search results from LinkedIn, Indeed, and other public websites (SKILL.md).
Boundary markers: The instructions do not define delimiters or specific 'ignore' warnings for the data retrieved via web_search.
Capability inventory: The agent has the ability to perform web searches, write to the local filesystem (/tmp/), and execute shell commands (rclone).
Sanitization: There is no evidence of sanitization or validation of the content retrieved from external websites before it is processed or used in outreach drafts.

ai-consultant