The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Python script (scripts/last30days.py) and various CLI tools (node, python3, yt-dlp, gh, xurl) to perform its primary function. It also executes standard system utilities like find, mktemp, and mktemp for cache management and temporary file handling.
[EXTERNAL_DOWNLOADS]: The skill connects to several well-known and legitimate APIs and services to gather research data, including:
api.scrapecreators.com (Social media scraping)
api.openai.com (Reddit discovery fallback)
api.x.ai and Twitter/X GraphQL APIs
hn.algolia.com (Hacker News)
gamma-api.polymarket.com (Polymarket)
Brave Search, Parallel AI, or OpenRouter for supplemental web results.
[DATA_EXFILTRATION]: No evidence of unauthorized data exfiltration. All network requests are directed towards official API endpoints or well-known research sources as part of the skill's documented primary purpose.
[PROMPT_INJECTION]: The skill contains complex instructions to the agent to avoid common "slop" or hallucination patterns, enforcing a specific "contract" for output (e.g., forbidding em-dashes and specific title formats), which serves as a defensive prompt engineering measure rather than a malicious injection.
[CREDENTIALS_UNSAFE]: The skill manages credentials via environment variables (e.g., SCRAPECREATORS_API_KEY, OPENAI_API_KEY) and specifically instructs users to store them in .env files, following industry best practices for secret management. No hardcoded secrets were detected.

last30days