The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill relies on dynamic code execution in multiple contexts: the run_sdk_snippet tool is used to execute Python code for data analysis and tool orchestration, and the create_slide tool accepts an analysis_js parameter which executes arbitrary JavaScript within a browser environment to perform layout quality checks.
[COMMAND_EXECUTION]: The skill instructs the agent to use shell_exec for file management operations, specifically using mv for renaming/moving slides and cp for copying user-uploaded images into the project directory.
[EXTERNAL_DOWNLOADS]: The skill fetches assets from well-known services and trusted CDNs, including TailwindCSS (via JIT compilation script), FontAwesome, Google Fonts, and ECharts. These are standard resources for web-based slide creation.
[PROMPT_INJECTION]: The skill's frontmatter contains a directive that attempts to override the agent's default skill selection behavior by mandating that this specific skill MUST be loaded whenever a specific pattern ([@slide_project:...]) is detected in the user's message.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection.
Ingestion points: The skill ingests untrusted data through web_search, read_webpages_as_markdown, and image_search tools.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands in the ingested data.
Capability inventory: The agent has access to run_sdk_snippet (Python execution), shell_exec (file system operations), and create_slide (JavaScript execution).
Sanitization: No sanitization or validation of the external content is mentioned before it is processed or used in project creation.

creating-slides