The Agent Skills Directory

[SAFE]: The skill package is a documentation-only resource for the Arashi worktree orchestration tool. It does not include executable scripts or third-party binary downloads within the package itself.
[SAFE]: The skill demonstrates high security awareness by explicitly instructing the AI agent to use structured --json output for all automation tasks, which prevents data misinterpretation or accidental execution of human-targeted text.
[SAFE]: It provides robust defenses against command injection by mandating that user-supplied paths and labels be passed as distinct process arguments rather than being interpolated directly into shell or script sources.
[SAFE]: The documentation includes explicit security boundaries, advising agents never to enter secrets (API keys, tokens, or passwords) into lifecycle hooks, configuration snippets, or interactive prompts.
[SAFE]: The skill emphasizes 'least privilege' and 'safe defaults,' recommending the use of --dry-run to preview destructive mutations and limiting the scope of network-heavy or broad filesystem operations via repository groups or specific filters.
[SAFE]: External resource references, such as the official documentation site (arashi.haphazard.dev) and the GitHub repository, are consistent with the author's identity (corwinm) and are documented neutrally as standard project resources.

arashi