The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the pbcopy command on macOS to copy generated prompts to the system clipboard (SKILL.md, Step 7). While functional, this involves direct interaction with system-level utilities.\n- [COMMAND_EXECUTION]: The workflow includes a file-writing step where the output path output/illustration/{タイトル}.md is constructed using a user-provided title (SKILL.md, Step 9). This creates a risk of path traversal if the user provides a title containing relative path sequences like ../, potentially allowing files to be written outside the intended directory.\n- [PROMPT_INJECTION]: The skill presents a vulnerability surface for indirect prompt injection as it processes untrusted user input and interpolates it into prompts and file operations without sanitization.\n
Ingestion points: User input is collected via the AskUserQuestion tool for six different prompt elements (SKILL.md, Step 3).\n
Boundary markers: No boundary markers or 'ignore embedded instructions' warnings are used when inserting user input into the PROMPT_TEMPLATE.md.\n
Capability inventory: The skill has the capability to write to the local file system and execute the pbcopy command.\n
Sanitization: There is no evidence of sanitization, validation, or escaping of the input provided by the user before it is used in file paths or generated instructions.

creating-illustrations