The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the mv shell command to migrate existing user project folders into a new directory structure (~/Desktop/OS/). This process is transparent, providing a migration plan and requiring user confirmation before execution.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests content from external sources (Sent emails, Slack messages, Notion, and Google Drive) to build a "Voice DNA" profile without using boundary markers or sanitization. This allows malicious instructions embedded in those sources to potentially influence the agent's behavior when generating context files.
[EXTERNAL_DOWNLOADS]: The skill points users to the official Claude Desktop download site (claude.ai/download) to enable tool integrations. This is a reference to a well-known, trusted service and does not involve executable downloads from unknown sources.
[PROMPT_INJECTION]: The skill sets up a self-correcting rules engine in the CLAUDE.md file, instructing the agent to dynamically append and prioritize new rules based on future interactions. While intended for workspace personalization, this creates a mechanism for persistent instruction modification driven by runtime data.

onboard