The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection.
Ingestion points: The agent ingests a user-provided topic in Step 1, which is then interpolated into a detailed prompt for the KIE AI model in Step 2 of SKILL.md.
Boundary markers: The instructions lack delimiters or instructions to ignore embedded commands when processing the user topic.
Capability inventory: The skill allows the agent to execute shell commands, perform network requests via curl, and run Python scripts (python), which could be misused if the generation prompt is successfully hijacked.
Sanitization: There is no logic provided to sanitize or validate the user topic before prompt construction.
[COMMAND_EXECUTION]: The skill performs command-line operations and script execution.
Evidence: Executes a local Python script scripts/infographic-builder/overlay_logo.py to composite logos onto images.
Evidence: Uses shell commands (grep, cut, source) to manage the KIE_API_KEY from a local .env file.
[EXTERNAL_DOWNLOADS]: The skill interacts with an external service to fetch assets and dependencies.
Evidence: Communicates with the KIE AI API (api.kie.ai) to generate and download PNG infographics.
Evidence: Recommends installing the Pillow library via pip, which is a standard image processing dependency.

infographic-builder