hokipoki

What This Skill Does

The HokiPoki skill enables seamless interoperability between different LLM providers directly from your development environment. By leveraging the HokiPoki P2P network, this tool allows you to route complex programming and analytical tasks to Claude, Codex, or Gemini without ever leaving your terminal or switching browser tabs. It acts as a specialized bridge, ensuring that your API keys remain local to your machine, while only the specific task instructions and necessary file context are securely exchanged with chosen providers. This is ideal for developers who want to compare outputs, leverage the unique strengths of various models, or distribute high-compute tasks across a team workspace.

Installation

To integrate HokiPoki, ensure you have Node.js installed on your system. Run the following commands in your terminal to set up the CLI and authenticate your session:

1. Install the global package: npm install -g @next-halo/hokipoki-cli
2. Authenticate: hokipoki login
3. Verify identity: hokipoki whoami

If you intend to host models for others, ensure Docker is installed and running, then execute hokipoki register --as-provider --tools <list> to begin contributing to the network.

Use Cases

• Second Opinions: If a current model is producing errors, instantly route the same files to another engine to debug or refactor.
• Task Delegation: Offload large-scale refactoring or documentation tasks to specific models known for those strengths (e.g., Gemini for massive context).
• Team Collaboration: Utilize shared team workspaces to allow colleagues to process tasks using your authorized API access in a secure, isolated manner.
• Environment Consistency: Maintain a consistent development workflow by keeping the CLI as your primary interface for all AI interactions.

Example Prompts

1. "The current model keeps failing on this regex; can we hop to Claude to see if it handles the logic better?"
2. "Hokipoki, please take the contents of this service directory and use Codex to add unit tests for the login function."
3. "I need a fresh perspective on this architectural design, ask another AI to review the security issues in my current project."

Tips & Limitations

• Security: HokiPoki utilizes local execution for provider mode, meaning code runs in isolated Docker containers. Always review patches before final commitment.
• Performance: When requesting help, using the --json flag is highly recommended for structured data handling and easier integration with automated build pipelines.
• Constraints: If auto-apply fails for git patches, ensure your directory is initialized with git init. Use --no-auto-apply if you prefer to review code changes manually before they are written to disk.