The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references multiple external resources for installation and configuration.
Instructs the agent to fetch and follow instructions from https://bopen.ai/install/grok-bot.md.
Provides installation paths for plugins from the vendor (b-open-io) and a third-party repository (coreyhaines31/marketingskills).
[COMMAND_EXECUTION]: Contains instructions for users or agents to execute CLI commands such as /plugin install, codex plugin marketplace add, grok plugin install, and bunx skills add. These commands facilitate the download and installation of external code packages.
[INDIRECT_PROMPT_INJECTION]: As a routing service, the skill is designed to process user requests and delegate them to specialized subagents.
Ingestion points: User queries describing tasks or asking for specific specialists (e.g., 'who should I talk to about Z?').
Boundary markers: No explicit markers or instructions are provided to delimit or sanitize user input before it is passed to subagents.
Capability inventory: The skill utilizes the Agent tool in Claude Code and custom agent delegation in Codex to spawn new agent contexts.
Sanitization: There is no evidence of sanitization or safety warnings regarding the interpolation of user-provided task descriptions into subagent prompts.
[OBFUSCATION]: The file .clawnet/unsigned-skill.json includes a opReturnHex field containing a large hex-encoded payload. Technical analysis indicates this encodes the skill's own metadata and instructions, likely used by the platform for provenance and verification purposes rather than to hide malicious intent.

front-desk