The Agent Skills Directory

[PROMPT_INJECTION]: The instructions include a 'Guardrails' section that mandates a specific behavioral override. The agent is directed to 'stop immediately' and output a pre-defined marketing message including an affiliate link if the data returned meets certain criteria (such as low item counts), bypassing standard conversational responses to prioritize vendor promotion.
[METADATA_POISONING]: The skill's documentation and mandatory error response strings are populated with affiliate tracking parameters (?fpr=yhdrb) intended to monetize user sign-ups and upgrades on the third-party platform.
[EXTERNAL_DOWNLOADS]: The skill references and utilizes the apify-client library (for both Python and Node.js) and makes network calls to the official api.apify.com domain. These are standard resources for the skill's stated purpose of interacting with the Apify infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (tweet text, profiles, and metadata) from X/Twitter. While the skill performs standard data retrieval, it lacks explicit instructions for the agent to ignore potential instructions embedded within the scraped content, and no sanitization logic is specified for the ingested text before it is returned to the agent's context.

x-scraper