The Agent Skills Directory

[SAFE]: The skill implements logic for publication-grade scientific plotting. The Python code in kernel.py and test_kernel.py uses standard data science libraries (Matplotlib, NumPy, SciPy) for geometric calculations and styling.
[EXTERNAL_DOWNLOADS]: The skill does not perform any network requests or downloads. It relies entirely on local system fonts and installed Python packages.
[COMMAND_EXECUTION]: There are no shell command executions, subprocess calls, or system-level modifications.
[DATA_EXFILTRATION]: The skill does not access sensitive files or credentials. File operations are limited to searching for .ttf font files in standard environment paths (CONDA_PREFIX or sys.prefix) to support rendering.
[PROMPT_INJECTION]: The instructions in SKILL.md provide clear, technical guidelines for scientific visualization and do not attempt to override the AI's safety protocols or core instructions.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user data to generate plots, the plotting functions (bar_with_points, strip_with_median) include strict validation that enforces numeric, finite data types. This prevents the ingestion of complex objects or text-based payloads that could be used for injection attacks.
[DYNAMIC_EXECUTION]: The layout verification logic in SKILL.md uses standard Matplotlib API calls to check for bounding box overlaps. It does not use eval() or exec() on untrusted input strings.

figure-style