The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python scripts (scripts/setup_deps.py, tests/verify-chart-pipeline.py, and scripts/generate_chart.py) that utilize the subprocess module. These calls are employed for legitimate administrative and testing purposes, such as creating virtual environments, installing the matplotlib and numpy packages via pip, and verifying the integrity of the chart generation pipeline.
[DATA_EXFILTRATION]: The skill is designed to analyze internal resources, including local codebases, documentation, and BI reports, to establish a research baseline. While this involves reading local data, the instructions are focused on generating internal research reports and benchmarking against external findings. No patterns for unauthorized data exfiltration to external third-party servers were detected.
[INDIRECT_PROMPT_INJECTION]: As the skill fetches and processes content from external websites via WebFetch and Playwright, it is theoretically exposed to indirect prompt injection from malicious web pages. However, the skill implements a mitigation strategy by requiring sub-agents to summarize and structure external data into compressed JSON formats before it is processed by the Lead Agent, reducing the risk of raw injection payloads affecting the core logic.
[EXTERNAL_DOWNLOADS]: The dependency setup script (scripts/setup_deps.py) fetches the standard matplotlib and numpy libraries from the official Python Package Index (PyPI).

deep-research