The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/illo.py engine interacts with local installations of the codex and grok command-line tools using subprocess.run. These calls are a primary feature used to leverage host-side image generation capabilities when available. The commands use absolute paths and pass data through standard input to minimize risks associated with shell interpolation.
[EXTERNAL_DOWNLOADS]: The skill includes a scripts/repair-hermes-assets.sh shell script intended for fixing corrupted binary assets (character model sheets). It fetches files from the skill's official GitHub repository (raw.githubusercontent.com/tmchow/illo-skill) and validates them using SHA256 checksums before overwriting any local assets. This process is restricted to the skill's own assets and uses a trusted source.
[SAFE]: The skill demonstrates robust security awareness by guiding the user to initialize API keys through a secure prompt that writes to a configuration file with restricted (600) permissions. The code intentionally avoids reading secrets from environment variables and includes a redact function to prevent sensitive tokens from appearing in subprocess output or error messages.

illo