The Agent Skills Directory

[CREDENTIALS_UNSAFE]: Hardcoded credentials found in multiple documentation files:
SKILL.md and references/alpaca-paper-trading.md contain an Alpaca Secret Key: 8mKU43vRGzgM3QFLwdTex1n3Li5FgEGehyVxJJi5Ags1.
references/alphavantage-api.md contains a hardcoded Alpha Vantage API Key: V03ZKLQR9FW5N7VF.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of several Python packages from public registries without specifying versions:
alpaca-py (official Alpaca SDK).
akshare (financial data library).
yfinance (Yahoo Finance wrapper).
parsimony-finnhub (An unofficial wrapper for the Finnhub API; the official package is typically finnhub-python).
[COMMAND_EXECUTION]: The skill instructions involve executing local Python scripts (us_stock.py) from a specific directory using a virtual environment, allowing for shell command execution as part of its operational workflow.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its processing of untrusted data from external sources.
Ingestion points: The skill fetches company news from Finnhub (/company-news), news sentiment data from Alpha Vantage (NEWS_SENTIMENT), and stock constituent information from Wikipedia.
Boundary markers: There are no instructions for using delimiters or boundary markers to isolate external data from the agent's core instructions.
Capability inventory: The agent has the capability to execute internal scripts and make network requests to various external domains.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from external news sources before it is processed by the agent.

us-stock-data