The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to clone and install code from a personal GitHub repository (simonlin1212/tradingagents-astock) which is not associated with a recognized organization.
[REMOTE_CODE_EXECUTION]: Installation involves running pip install -e . on a repository from an unverified source. This can result in the execution of arbitrary code via the package's setup scripts (e.g., setup.py) during the installation process.
[COMMAND_EXECUTION]: The framework facilitates the execution of shell commands for its core functionality, including git operations and custom CLI entry points like tradingagents and tradingagents-web.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by ingesting unstructured data from external news feeds (CLS.cn, EastMoney, Sina) and social sentiment sources.
Ingestion points: Untrusted data enters the agent context through tools such as get_news, get_social_sentiment, and get_global_news across various analyst files.
Boundary markers: The instructions do not define delimiters or specific warnings to ignore embedded instructions within the ingested financial content.
Capability inventory: The skill has significant capabilities, as the agents generate trading signals (BUY/SELL) and execution plans which are used for portfolio management decisions.
Sanitization: There is no evidence of sanitization or validation of the external content before it is interpolated into the analyst agents' prompts.

tradingagents-astock-multi-agent-framework