The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the infsh command-line utility to interact with backend services, including user authentication and running specific applications as part of the visualization workflow.
[EXTERNAL_DOWNLOADS]: The documentation references an external URL on GitHub (https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md) to provide users with platform installation instructions, which is standard for platform-specific skills.
[REMOTE_CODE_EXECUTION]: The skill utilizes the infsh/python-executor and infsh/html-to-image tools to execute code and render visualizations remotely. This behavior is expected as it is the core functionality for generating charts and KPI cards within the platform environment.
[PROMPT_INJECTION]: The skill contains patterns where user data is interpolated into code templates for execution. This creates a surface for indirect prompt injection if an agent processes untrusted data without validation before placing it into the Python script strings.
Ingestion points: Data for chart categories, values, and labels processed in SKILL.md templates.
Boundary markers: None present in the code templates to separate data from code.
Capability inventory: Python code execution via infsh/python-executor and file creation capabilities.
Sanitization: No explicit sanitization or validation logic is provided in the chart recipes.

data-visualization