The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command python3 scripts/split_sprite_sheet.py with multiple parameters derived from dynamic variables such as <source.png>, <output-dir>, and <pack-id>. The direct interpolation of these variables into a shell string without explicit sanitization or quoting poses a risk of command injection if the inputs are influenced by untrusted data.
[REMOTE_CODE_EXECUTION]: The skill depends on a Python script at scripts/split_sprite_sheet.py. This script is not provided in the analyzed file set, and its execution constitutes a reliance on unverified code logic.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through its ingestion of external assets and configuration files.
Ingestion points: Reads data from contracts/style-contract.yaml, contracts/component-contract.yaml, mappings/*.yaml, and user-provided image files in SKILL.md.
Boundary markers: No delimiters or safety instructions are provided to prevent the agent from following malicious instructions potentially embedded in these data sources.
Capability inventory: The agent can perform file system writes, zip archive creation, and shell command execution via python3.
Sanitization: There are no instructions for sanitizing or validating the content of the ingested YAML files or the metadata extracted from images.
[DATA_EXFILTRATION]: The skill creates a ZIP archive at <project>/packages/<pack-id>-png.zip. This mechanism could potentially be manipulated to include sensitive project files in the archive if path variables or file selectors are not strictly validated.

game-ui-sprite-sheet-splitter