The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from external websites during its research phase, creating a surface for indirect prompt injection. If a target website contains malicious instructions, the agent may inadvertently follow them while generating content or executing its workflow.
Ingestion points: The analyze-web.js script scrapes content, brand data, and competitor info from user-provided URLs and internal subpages.
Boundary markers: There are no explicit delimiters or system instructions defined to isolate the scraped content from the agent's core logic.
Capability inventory: The skill can generate images via Gemini, write to local JSON files, and publish posts to TikTok and Instagram via the Upload-Post API.
Sanitization: No validation or sanitization of the scraped website content is specified before it is fed into the generative models.
[COMMAND_EXECUTION]: The skill maintains autonomous persistence by programmatically managing its own execution schedule via system utilities.
Evidence: The skill is designed to 'self-schedule' by reading analytics data and setting its next execution time using cron to ensure daily automated posting without human intervention.
[EXTERNAL_DOWNLOADS]: The skill downloads external browser binaries required for its web analysis capabilities.
Evidence: The skill executes playwright install chromium to acquire the necessary browser engine for JavaScript-rendered web scraping.

marketing-carousel-growth-engine