The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed around the execution of the image-mcp CLI tool to perform image-related tasks. It includes instructions for running various subcommands such as create, edit, login, and doctor to manage the image generation lifecycle.
[DATA_EXFILTRATION]: The skill documentation notes that local file paths provided to the image-mcp edit command are automatically uploaded to the Image MCP service. This is a standard functional requirement for remote image editing services and is clearly documented as part of the workflow.
[PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface by taking user-provided text and interpolating it into CLI commands (e.g., image-mcp create "<prompt>").
Ingestion points: User-provided generation prompts and local file paths/URLs in SKILL.md and references/cli-usage.md.
Boundary markers: The instructions recommend wrapping prompts in double quotes within the CLI commands.
Capability inventory: The image-mcp CLI tool possesses network communication capabilities and the ability to read local files for uploading.
Sanitization: There are no explicit instructions for the agent to sanitize or validate the content of the user prompts before passing them to the CLI.

image-mcp