The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate image prompts, which are then interpolated into shell commands for local execution.
Ingestion points: User-provided text descriptions (e.g., "crie uma imagem de X") are ingested as described in the workflow sections of SKILL.md.
Boundary markers: Absent. The instructions do not define delimiters or warnings to prevent the agent from executing embedded instructions within the user prompt.
Capability inventory: The skill utilizes shell command execution via python generate.py with multiple flags.
Sanitization: Absent. There is no guidance for the agent to escape shell metacharacters or validate the input before constructing the command line.
[COMMAND_EXECUTION]: The skill defines a workflow that relies on the execution of local scripts (generate.py) with parameters influenced by the user.
Evidence: SKILL.md contains multiple examples of shell commands, such as python generate.py --template "instagram-lifestyle" --customization "..." and python generate.py generate --prompt "...".

image-studio