project-init

You can set up GitHub Projects for AI agents through a guided interview that creates custom fields, labels, config blocks, and canonical files in one session. This ensures all project management infrastructure is wired together correctly without manual configuration.

A CLAUDE.md config block provides AI agents with project context and operational instructions within your repository. It is required to establish GitHub as the business operations hub, and this guided setup writes or appends to it without overwriting existing content.

Yes, the guided setup supports multi-repo configurations. It gathers repository scope and business context via interview prompts, then writes or appends to AGENTS.md or CLAUDE.md across the relevant repositories while avoiding overwrites.

Custom fields and labels for retro and planning cycles are created through a guided interview workflow. It automatically generates a GitHub Project with fields like Status, Deadline, and Week, plus the necessary labels for project management cycles.

No, the setup intentionally avoids overwrites. When creating canonical files and config blocks, it safely writes or appends new content to existing AGENTS.md or CLAUDE.md files without destroying your current repository configuration.

You need a Claude skills directory to copy the skill into, and an existing GitHub repository scope. The guided interview then handles the rest, gathering business context to generate the necessary project management infrastructure.