The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script scripts/_skill_cli.py includes a run_command function that executes shell commands via subprocess.run. This is intended for running and verifying the CLI tools designed by the agent and incorporates basic timeout and error handling.
[DYNAMIC_EXECUTION]: The skill instructions in SKILL.md prompt the agent to execute a local vendor-provided script, suggest_next_skill.py, upon completion of the primary task to determine the next workflow step. This is a controlled execution of the skill's own components.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of @clack/prompts, consola, and picocolors via npm. These are well-known, trusted, and widely-used libraries for creating interactive terminal user interfaces.
[INDIRECT_PROMPT_INJECTION]: The skill's verification process involves the agent reading and analyzing the stdout of the scripts it creates, which presents a theoretical injection surface for data processed during verification.
Ingestion points: Terminal stdout from executed CLI scripts (SKILL.md).
Boundary markers: No explicit delimiters are defined for the captured stdout.
Capability inventory: Command execution via subprocess.run (scripts/_skill_cli.py) and npm package installation.
Sanitization: The agent is instructed to check alignment and borders manually, but no automated sanitization of the terminal output is implemented.

cli-interface-design