The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts scripts/auto_generate_demo.py and scripts/batch_generate.py facilitate arbitrary command execution. They take user-provided strings and incorporate them into .tape files which are then executed by the vhs binary via subprocess.run. This allows any command passed to the skill to run in a real shell environment.
[EXTERNAL_DOWNLOADS]: The skill requires and provides instructions for installing external software dependencies, specifically vhs and asciinema. While these are well-known tools, the skill's reliance on external binaries from GitHub and third-party package managers (brew, apt, go) is a necessary component of its functionality.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by processing external data and executing commands based on that data.
Ingestion points: scripts/batch_generate.py reads demo definitions from YAML and JSON configuration files, while scripts/auto_generate_demo.py takes command strings as input arguments.
Boundary markers: Absent. Commands provided in the input data are directly interpolated into the executable .tape files without sanitization or instructions to ignore embedded malicious content.
Capability inventory: The skill has the capability to execute any shell command on the host system via the vhs recording tool.
Sanitization: No input validation, escaping, or filtering is performed on the commands before they are executed in the recording session.

cli-demo-generator