cursor-kenji

cursor-kenji is an open-source playbook pack designed for AI coding tools including Cursor, Claude Code, Codex CLI and Gemini CLI. It provides 145 agent skills, 57 slash commands, 6 background subagents, 16 MCP server templates, 12 Cursor-specific IDE skills, 1 completion hook and 7 drop-in project rules, tuned for React/Next.js/Supabase stacks while working with almost any tech stack. Users trigger pre-defined workflows by describing tasks in natural language, instead of memorizing command names. Built-in playbooks cover the full development lifecycle: onboarding to a new codebase, pre-build requirement alignment, feature development (spec → failing test → code → smoke test → PR), bug fixing and shipping, security audits (OWASP-style findings with file:line references), production readiness checks, post-deployment observation and rollback, among others. Skills can be combined for complex workflows, such as pairing exploratory monkey testing with Playwright regression passes. It is installed via a single CLI command: `npx @kensaurus/cursor-kenji --all` installs all assets to all supported tools at once, with flags to target specific tools (Cursor, Claude Code, Codex CLI, Gemini CLI), preview changes, verify installed file hashes, or install only specific skills. For Claude Code, it is also available as a plugin via the built-in marketplace. After installation, users can configure optional MCP servers (for Supabase, GitHub, Playwright, AWS, Slack etc.) by copying a template config file and setting required environment variables. All skills follow the Agent Skills specification, and the package includes automated tests to validate skill counts, spec compliance and install behavior. It is not a hosted service or a Cursor replacement, but a local, installable set of guardrails and automation playbooks to reduce repetitive work for developers using AI coding assistants.