The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions guide the agent to execute a local setup script (./omw) or the CLI entry point (node lib/cli.js). This is standard behavior for infrastructure setup and installation tasks.
[INDIRECT_PROMPT_INJECTION]: The skill manages and merges content within the AGENTS.md file, which may contain user-provided instructions.
Ingestion points: Reads AGENTS.md from the project root during the setup process.
Boundary markers: The skill explicitly uses markers ( and ) to delimit managed content, which helps prevent the agent from misinterpreting user-authored text as system instructions.
Capability inventory: The skill performs directory creation and file writes in the project and user home directories (SKILL.md).
Sanitization: No explicit sanitization of existing AGENTS.md content is described; however, the use of structural markers is a recognized safety practice for managing shared configuration files.

omx-setup