The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for directory creation (mkdir), template interpolation (sed), and symbolic link creation (ln -sf) to install skills in platform configuration directories such as ~/.claude/skills and ~/.copilot/skills.
[INDIRECT_PROMPT_INJECTION]: The skill ingests free-form user input during the brainstorming phase and interpolates these values into the generated SKILL.md file using sed. 1) Ingestion points: User-provided skill purpose and descriptions. 2) Boundary markers: None present in the templates to delimit user input from instructions. 3) Capability inventory: File system writes, shell script execution, and symbolic link creation. 4) Sanitization: No sanitization or escaping is performed on user input before interpolation.
[DATA_EXFILTRATION]: The skill retrieves the user's name and email by executing git config user.name and git config user.email for use in generated metadata.
[PRIVILEGE_ESCALATION]: The scripts/init_skill.py script modifies file permissions using chmod(0o755) for dynamically created scripts within the new skill directory.

skill-creator