The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses tmux to manage persistent background sessions for AI agents using the claude --remote-control command. It also utilizes ps for process verification and the open command on macOS to display the generated report.
[EXTERNAL_DOWNLOADS]: The skill instructs agents to search for and retrieve research data and examples from external sources including GitHub and X/Twitter.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8) due to its processing of repository files and external research content. 1. Ingestion points: Repository content and external research data. 2. Boundary markers: Absent. 3. Capability inventory: Background shell access and task dispatching. 4. Sanitization: None specified. This risk is managed by the skill's read-only analysis design and human-gated execution.

night-shift