bkit-gemini and bkit-codex
These are ecosystem siblings offering parallel implementations of the same PDCA + AI-native development framework—one optimized for Gemini's API and the other for OpenAI's Codex—allowing developers to choose their preferred LLM backend while maintaining consistent vibecoding methodology.
About bkit-gemini
popup-studio-ai/bkit-gemini
bkit Vibecoding Kit for Gemini CLI - PDCA + AI-native development
Implements a 10-event hook system that intercepts Gemini CLI's full lifecycle to enable phase-aware automation—including intent detection across 8 languages, phase-specific tool filtering, and automatic PDCA state transitions—without modifying the CLI source. Organizes 35 domain skills and 21 specialized agents (CTO, architects, product managers, QA) into a context engineering framework that progressively discloses knowledge and enforces behavioral rules through TOML commands and markdown modules, with built-in permission controls that block dangerous commands during planning phases.
About bkit-codex
popup-studio-ai/bkit-codex
bkit Vibecoding Kit for OpenAI Codex - PDCA + Context Engineering for AI-native development
Implements a three-tier context management system with 27 skills, 16 MCP tools, and a 9-phase development pipeline that enforces PDCA methodology through instruction-driven guidance rather than hook-based automation. The architecture uses progressive disclosure to minimize context token usage—loading skill metadata first, full content on-demand, and reference files only when explicitly requested. Built as a zero-dependency Node.js MCP server with automatic state recovery, multi-language support, and persistent session memory across Codex CLI invocations.
Scores updated daily from GitHub, PyPI, and npm data. How scores work