# vibecode-pro-max-kit

A spec-driven coding harness that adds persistent context memory, 15 agents and 33 skills to Claude Code and Codex, running a 7-phase plan-first workflow with self-healing check loops.

vibecode-pro-max-kit drops a plan-first development process into any project for Claude Code, Codex, Cursor, Windsurf, Copilot, or another AI coding agent, working with any tech stack or language. It runs a seven-phase workflow (Research, Spec, Innovate, Plan, Validate, Execute, Update-Process) that stops the agent from jumping straight to code, plus an autopilot mode that runs several phases in a row without stopping and can resume in a fresh session using a saved `/goal` token.

Self-healing plan-check-fix and test-check-fix loops look for gaps and fix them automatically across up to ten cycles each, feasibility probes give a viable/not-viable verdict before a design is committed to, and a strategy picker weighs one agent, several agents, or a coordinated team against estimated cost before each phase. Large projects are split into independent phases with quality gates between them, and 36 validators check the kit's own structure for drift.

It installs with a single curl command that detects new versus returning users and never overwrites existing files; it is released under the MIT license.

Website: https://github.com/withkynam/vibecode-pro-max-kit