The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external data to generate plot labels and titles, which represents a potential attack surface for indirect prompt injection if the source data is malicious.\n
Ingestion points: The helper functions in kernel.py (e.g., bar_with_points, strip_with_median, focal_palette) accept labels and numeric values from external datasets.\n
Boundary markers: No explicit delimiters or boundary markers are defined for text labels derived from data.\n
Capability inventory: The skill is capable of writing files to the local file system using the savefig and save methods.\n
Sanitization: No explicit sanitization or filtering is performed on data-derived strings before they are rendered into the figure context.\n
[DYNAMIC_EXECUTION]: The skill's documentation suggests executing local code from kernel.py to initialize the environment.\n
Evidence: The SKILL.md file (§Dependencies) instructs the user or agent to load the helper module via exec(open("kernel.py").read()).\n
Context: This is a routine initialization pattern for loading local script logic in interactive coding environments and is considered a low-risk operational finding.

figure-style