The Agent Skills Directory

[SAFE]: The skill demonstrates best-in-class credential hygiene. It explicitly instructs the user to store the Admin API token in a local .env file instead of pasting it into the chat. The internal Python engine reads this token directly, ensuring that sensitive credentials never enter the LLM context or shell command history.
[SAFE]: All write operations are protected by a robust approval mechanism. The tool generates a plan, calculates its SHA-256 hash, and requires the user to provide this exact hash along with a --yes flag before any mutation occurs on the remote store. This prevents the AI agent from making unintended changes without explicit user verification of the exact execution plan.
[SAFE]: The skill enforces secure local data practices. The 'run-manifest.json' file, which contains the created campaign's API key, is written using atomic file operations and restricted permissions (chmod 600 on supporting systems). This ensures that the secret remains private to the current user.
[SAFE]: Network security is strictly enforced via a custom HTTP client in 'scripts/campaign_admin.py' that implements same-origin validation. It refuses to follow redirects and ensures that authentication tokens are only sent to the validated subdomain of the official platform (29next.store).
[SAFE]: Supply chain risk is minimized by avoiding external dependencies. The Python scripts rely exclusively on the standard library, and the update mechanism is a read-only version check against the author's official GitHub repository.

next-campaigns-create