The Agent Skills Directory

[Data Exposure] (SAFE): The script scripts/test_institutional_endpoint.py retrieves the FMP_API_KEY from environment variables using os.environ.get(), avoiding hardcoded credentials.
[External Network Access] (LOW): The skill communicates with finviz.com and financialmodelingprep.com. While these are non-whitelisted domains, the network activity is essential for the skill's primary purpose of fetching stock market data.
[Indirect Prompt Injection] (LOW): The skill ingests untrusted data from external web pages and APIs.
Ingestion points: scripts/finviz_stock_client.py fetches HTML from Finviz; scripts/test_institutional_endpoint.py fetches JSON from FMP.
Boundary markers: Absent; the data is directly parsed into dictionaries.
Capability inventory: Limited to calculating numeric scores and returning structured data; no shell execution or file-writing capabilities were detected.
Sanitization: The skill uses BeautifulSoup to target specific table elements (snapshot-table2), which effectively filters out most unrelated page content.
[Unverifiable Dependencies] (SAFE): The skill requires standard, well-known libraries (beautifulsoup4, requests, lxml) which are common for Python data processing and do not involve remote code execution during installation.

canslim-screener