The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly bypass standard agent safety and oversight protocols. It directs the agent to 'never ask for permission between steps' and maintain 'Zero Confirmation' throughout the pipeline, only notifying the user at the end of the process.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection.
Ingestion points: The analyze-web.js script uses Playwright to scrape content from any user-provided website URL, including internal pages like pricing and testimonials.
Boundary markers: There are no specific instructions or delimiters provided to help the agent distinguish between its core instructions and potentially malicious content embedded in the target websites.
Capability inventory: The skill has the ability to publish content to external APIs (publish-carousel.sh), generate images (generate-slides.sh), and modify system scheduling (cron).
Sanitization: No sanitization or validation logic is mentioned for the content extracted from websites before it is used to drive image generation and social media captions.
[PERSISTENCE]: The skill attempts to maintain a continuous presence by self-scheduling its own execution. It instructs the agent to read performance data and 'set next cron execution at the optimal hour,' which establishes a persistence mechanism on the host system.
[COMMAND_EXECUTION]: The skill relies on several shell scripts (generate-slides.sh, publish-carousel.sh, check-analytics.sh) and external tools like Playwright to perform its tasks, requiring the agent to execute multiple system-level commands and manage browser dependencies like Chromium.

agency-carousel-growth-engine