The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) and Git to automate repository management tasks, including forking, cloning, and pushing. These actions are well-defined, core to the skill's function, and are performed within the scope of the user's authenticated GitHub environment.
[DATA_EXFILTRATION]: Contains a dedicated memory-cleansing procedure that deletes logs, state files, and issue trackers from the cloned repository before it is pushed to the fork. This safeguard ensures that sensitive activity history or parent-instance metadata is not propagated to child instances.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input through the ${var} parameter in SKILL.md to define the name and purpose of the new instance. This untrusted data is interpolated into documentation files (such as SETUP.md, MEMORY.md, and CLAUDE.md). The security risk is considered safe because the generated instance is created in an inert state, requiring the owner to manually add API keys and notification secrets before it can perform any autonomous operations.

spawn-instance