The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs several local command executions to fulfill its primary functions. It uses npx playwright screenshot to render dynamically generated HTML into image files. It also executes local Python scripts, specifically generate_image.py via uv run and an upload_image.py script located at an absolute path in the user's home directory.
[CREDENTIALS_UNSAFE]: The workflow instructions guide the agent to programmatically extract the GEMINI_API_KEY from a local environment file at ~/.claude/.env using the grep command. While necessary for the skill's operation, this involves accessing sensitive configuration files.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by interpolating untrusted user data into executable contexts.
Ingestion points: User-provided article content and paths are used to populate both AI prompts and HTML templates for rendering.
Boundary markers: No specific delimiters or instructions are used to separate user data from the system's HTML/CSS templates or prompt structures.
Capability inventory: The skill possesses the capability to execute shell commands (via Playwright and Python) and write files to the local file system.
Sanitization: There is no evidence of sanitization or escaping of user-provided strings before they are injected into HTML, which could allow for unintended rendering behavior.

huashu-wechat-image