The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill is designed to automatically install external dependencies using package managers if they are not detected in the environment. While the packages are well-known, automated installation without version pinning is a recognized security concern.
Evidence: SKILL.md states dependencies are automatically installed. scripts/build_pptx.js uses require.resolve to check for pptxgenjs, playwright, and sharp. scripts/read_excel.py and scripts/read_pptx.py check for pandas, openpyxl, python-pptx, and Pillow.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data files, which could theoretically contain malicious instructions intended to manipulate the agent's output or actions.
Ingestion points: scripts/read_excel.py and scripts/read_pptx.py are used to read user-supplied Excel and PowerPoint files.
Boundary markers: There are no explicit instructions or delimiters mentioned to prevent the agent from following natural language instructions found within the data cells or slides.
Capability inventory: The skill has the ability to execute shell commands (node, python, npx playwright) and write files to the local system.
Sanitization: No sanitization or filtering is performed on the content extracted from the data files before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic JavaScript execution to perform its visualization tasks.
Pattern: scripts/html2pptx.js uses Playwright's page.evaluate() to execute code inside a headless browser environment to extract DOM information and computed styles from HTML templates.
[COMMAND_EXECUTION]: The workflow relies on shell command execution for its primary features.
Patterns: The skill documentation in references/workflows.md and SKILL.md instructs the agent to run npx playwright screenshot, node build_pptx.js, and various Python scripts to handle data transformation and report generation.

huashu-data-pro