The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs a search for an executable script named tzai-image within standard hidden directories in the user's home folder (e.g., .agents, .claude, .cursor). Once found, it executes this script using bash to process image generation tasks.
[EXTERNAL_DOWNLOADS]: The skill identifies a dependency on the tzai-image-skill package. If the engine is missing, it suggests an installation command using npx pointing to the author's own repository (kedoupi/tzai-image-skill). This is standard behavior for modular skill architectures.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a prompt-engineering wrapper, taking user input and formatting it into a "visual brief" passed to the underlying engine. It documents an ingestion surface for untrusted data, but follows expected patterns for an image generation utility.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill references the use of an environment variable TZAI_API_KEY for authentication with the TaoziAPI. It does not contain hardcoded credentials or evidence of unauthorized data transmission.

tzai-logo