The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The blue launcher script dynamically fetches its core logic and dependencies from GitHub repositories belonging to the getcolors organization. These external resources are strictly pinned to specific, immutable Git commit hashes (SHAs) to ensure integrity and prevent supply chain tampering.
[COMMAND_EXECUTION]: The launcher script uses subprocess.run to execute the uv package manager and the Python interpreter. It forwards all command-line arguments from the user to the underlying deployment library, which is a standard pattern for CLI-based agent skills.
[PRIVILEGE_ESCALATION]: The skill documentation includes instructions for users to retrieve database credentials from sensitive file paths on remote machines using sudo via SSH. It also requires the user to grant execution permissions (chmod +x) to the local launcher script as part of the setup process.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted configuration data (e.g., colors.yml, pyproject.toml) and environment variables from the deployment repository. It possesses high-privilege capabilities including AWS management, Cloudflare DNS control, and remote command execution.
Ingestion points: Configuration files, environment variables, and repository structure in the user's workspace.
Boundary markers: Documentation emphasizes that generated artifacts in .colors/ are deterministic and should not be manually edited.
Capability inventory: Full cloud provider access (AWS, Cloudflare), infrastructure-as-code execution (OpenTofu, Ansible), and remote machine access (SSH, sudo).
Sanitization: Relies on pinned vendor-managed libraries for execution logic; no explicit input filtering is implemented in the primary launcher script.

package-neon-multi-node-blue