The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's fundamental operation involves scraping untrusted data from social media (Instagram, TikTok, X, Reddit) and arbitrary websites (via Website Content Crawler). This data is then returned to the AI for tasks such as sentiment analysis, summarization, and lead qualification.
Ingestion points: Data enters the agent's context through commands like apify datasets get-items and the outputs of various Actors (e.g., apify/instagram-scraper, apify/website-content-crawler), as seen in SKILL.md and the workflow guides in references/workflows/.
Boundary markers: There are no instructions to use specific delimiters (like XML tags) or to provide the model with 'ignore' instructions when processing the scraped content to prevent it from following embedded malicious commands.
Capability inventory: The skill utilizes the Apify CLI for network and compute operations (apify actors call, apify actors start) and mentions the Write tool for file system access.
Sanitization: The instructions do not describe any sanitization, filtering, or validation of the retrieved text before it is processed by the LLM.
[COMMAND_EXECUTION]: The skill makes extensive use of the apify-cli to interact with the Apify platform. It provides specific command patterns for starting actors, polling runs, and retrieving datasets, which are executed locally within the agent's environment.
[EXTERNAL_DOWNLOADS]: The skill explicitly instructs the agent to install the apify-cli tool globally using npm install -g apify-cli if it is not already present. This is a dependency on a public software registry and a third-party vendor (Apify).

apify-ultimate-scraper