The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a repository from an untrusted source: https://github.com/simonlin1212/tradingagents-astock.git.
[REMOTE_CODE_EXECUTION]: Following the clone, the skill directs the user to run pip install -e ., which executes the setup scripts or build backend of the untrusted repository, allowing for arbitrary code execution during installation.
[INDIRECT_PROMPT_INJECTION]: The framework ingests untrusted data from multiple external sources including news feeds, social sentiment, and various web-based financial portals.
Ingestion points: The get_news, get_global_news tools and the social_analyst agent process data from sources like EastMoney, Sina, and CLS.cn.
Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are documented for the data ingestion pipeline.
Capability inventory: The framework has the capability to generate trading signals (BUY, SELL, HOLD) and execute complex multi-agent analysis based on the ingested data.
Sanitization: There is no evidence of sanitization, filtering, or validation of the external content before it is interpolated into the agent prompts.
[COMMAND_EXECUTION]: The skill provides various CLI commands (tradingagents, tradingagents-web) that execute the installed framework's logic, which could trigger malicious behavior if the downloaded package has been tampered with.

tradingagents-astock-multi-agent-framework