The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions include the execution of shell commands to activate a virtual environment and run data-retrieval scripts.
Evidence: references/data-queries.md contains commands to activate a .venv in a parent directory (source ../../.venv/bin/activate) and execute a local Python script (python ../findata-toolkit-cn/scripts/views_runner.py).
[EXTERNAL_DOWNLOADS]: The skill performs external package installation as part of its environment setup process.
Evidence: references/data-queries.md instructs the agent to run python -m pip install -r ../findata-toolkit-cn/requirements.txt, which fetches packages from standard registries.
[INDIRECT_PROMPT_INJECTION]: The skill summarizes and analyzes data retrieved from external financial APIs and stock exchange sources, which could potentially contain malicious instructions.
Ingestion points: Market data is fetched via views_runner.py from sources like AKShare, the Shanghai Stock Exchange (SSE), and the Shenzhen Stock Exchange (SZSE), as described in references/methodology.md.
Boundary markers: The instructions do not define explicit delimiters or "ignore instructions" wrappers to isolate the retrieved data from the agent's logic.
Capability inventory: The skill has the capability to execute shell commands and Python scripts, which increases the potential impact if the agent follows instructions embedded in the external data.
Sanitization: There is no evidence of data sanitization, filtering, or validation of the external content before it is processed by the agent.

market-overview-dashboard