The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions and examples for executing shell commands (plotloom, cp) that incorporate user-controlled variables such as <slug> and <character-slug> as arguments. There is no evidence of sanitization or escaping for these inputs. A malicious user could provide an input containing shell metacharacters (e.g., ;, &&, |, or backticks) to execute arbitrary commands on the host system. This vulnerability surface is documented across the 'Plotloom Command Shape' and 'Workflow' sections.
Ingestion points: User-provided character slug and character concepts (SKILL.md).
Boundary markers: None present; user inputs are directly interpolated into shell command strings.
Capability inventory: Execution of the plotloom CLI tool and the cp command (SKILL.md).
Sanitization: No validation or sanitization logic is described for the variables used in command execution.
[DATA_EXPOSURE]: The skill uses a hardcoded local directory path (/Users/haha/plotloom_repo/) in its example commands. While likely intended as a placeholder, this practice exposes a specific local username and directory structure to the agent, which could lead to errors, information disclosure, or unintended file operations if the agent attempts to use the path literally on a different system environment.

plotloom-character-reference-sheet