The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains explicit instructions to bypass standard human-in-the-loop safety protocols and operate without user oversight.
Evidence: Instructions such as "Zero Confirmation: Run the entire pipeline without asking for user approval between steps", "Notify Only at End: The user sees results (published URLs), not process updates", and "You never ask for permission between steps — you research, generate, verify, publish, and learn".
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves processing untrusted data from external websites, which could contain instructions meant to subvert the agent's behavior.
Ingestion points: The analyze-web.js script uses Playwright to scrape content from target URLs, including internal pages like pricing and testimonials.
Boundary markers: The provided documentation does not specify the use of delimiters or instructions to ignore embedded commands in the scraped data.
Capability inventory: The agent can perform network requests (publishing via Upload-Post API), execute shell scripts (generate-slides.sh, publish-carousel.sh), and write to the local file system (learnings.json, analysis.json).
Sanitization: No sanitization or validation logic for the scraped content is mentioned.
[COMMAND_EXECUTION]: The skill's workflow relies on the execution of multiple shell, Python, and JavaScript scripts which are not fully disclosed in the main definition.
Evidence: References to generate-slides.sh, generate_image.py, publish-carousel.sh, check-analytics.sh, analyze-web.js, and learn-from-analytics.js.
[PERSISTENCE]: The skill describes a mechanism for self-scheduling and recurring execution, which can be used to maintain persistence on a system.
Evidence: "Self-Schedule: Read learnings.json bestTimes and schedule next execution at the optimal posting time" and "adjusts cron schedule so next execution happens at peak engagement hour".

Carousel Growth Engine