The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation provides instructions to install the omniroute package from the public npm registry using the command npm install -g omniroute or execute it directly using npx.
[COMMAND_EXECUTION]: The skill describes how to execute various subcommands of the omniroute tool, such as chat, repl, and stream, with support for flags that interact with the local file system (--file <path>) and standard input (--stdin).
[PROMPT_INJECTION]: The skill is designed to process user-supplied input and external file content, which represents a potential surface for indirect prompt injection.
Ingestion points: User prompts via command arguments, file content via the --file flag, and input via --stdin in the chat and stream subcommands.
Boundary markers: There are no markers or system instructions documented to delimit external data from agent instructions.
Capability inventory: The tool performs network communication with AI providers and file system reads; however, it does not involve subprocess execution of the retrieved content or persistence mechanisms.
Sanitization: No specific validation or sanitization of the input data before it is processed by the AI models is described in the provided documentation.

cli-chat