The Agent Skills Directory

[SAFE]: The skill serves as a thin wrapper to delegate complex system-building logic to a specialized subagent (meta-builder-agent). This modular design adheres to safe architectural patterns.
[COMMAND_EXECUTION]: The skill documentation includes shell-based logic for argument parsing and mentions the use of tools like git and jq. All variable handling in the provided logic is properly quoted, preventing command injection vulnerabilities during argument processing.
[DATA_EXFILTRATION]: There are no indications of network operations or attempts to access sensitive user credentials or private files. Its operations are restricted to the local .claude/ directory and related system configuration paths.
[PROMPT_INJECTION]: The skill does not contain any instructions that attempt to bypass AI safety guidelines, extract system prompts, or override core agent behavioral constraints.

skill-meta