claude-toolbox plugin
A configuration and kk plugin for Claude Code and Codex: skills from design to docs, multi-model review and shared memory
- 155 ★ on the repository
- Last commit
Extended access
Why this level
- The configuration changes permissions, hooks and MCP servers and multi-model review sends code to external providers
- The plugin connects external memory and documentation services and includes a protective action guard
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/claude-toolboxDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The configuration changes permissions, hooks and MCP servers and multi-model review sends code to external providers
- The plugin connects external memory and documentation services and includes a protective action guard
This is third-party code. Review the repository files before installing.
What it does
claude-toolbox is a set of settings and tools for Claude Code and Codex that the author describes as dotfiles for agents. It gives a minimal configured baseline (base permissions, a rich statusline with themes, MCP server wiring, sync across projects) and the `kk` plugin with sixteen development skills. Skills run as `/kk:<skill>` and form a pipeline: `/kk:design` (a design document, plan and task list), `/kk:review-design`, `/kk:implement`, `/kk:review-code`, `/kk:test`, `/kk:document`. Helpers exist too, such as `/kk:brainstorm` (an open conversation about an idea ending with a chat recap) and `/kk:clarify-docs` (clarifying issue descriptions and documents). Skills are language-aware: separate instructions for Go, Java, JS and TS, Kotlin, Kubernetes and Python. Code gets independent review from subagents and external models (Gemini and others), and memory of findings, decisions and conventions persists between sessions through Capy. Context7 pulls current library docs. There are three ways to use it: a repository template for a new project, adopting the settings into an existing one and installing only the plugin (`claude plugin install kk@claude-toolbox` after adding the marketplace).
Who it is for. Developers who work with Claude Code or Codex daily and want one pipeline from idea to documentation with reviews and shared memory.
Good fit when
- A new feature needs a design document, plan and task list and you want to refine it with questions
- Before writing code you want the design run through an independent review to find gaps
- Code review should be done by a separate subagent or an external model without the implementer involved
- The same Claude Code settings are needed across many projects with automatic sync
Not a fit when
- You need a single small skill: the configuration and plugin contain a whole set covering many scenarios
- You need a license with explicit terms: the repository formally has an undetermined license type
- You need other programming languages: detailed instructions exist for Go, Java, JS and TS, Kotlin, Kubernetes and Python
Example request
Run /kk:design for a CSV report export feature: ask questions, then write the design doc, plan and task listLimitations
The full configuration changes permission settings, hooks and MCP servers in your projects, and the `kk` plugin includes a protective guard for secrets and dangerous actions. Before adopting it in a working repository read the adoption section of the docs. Multi-model review calls external models, so part of the code goes to their providers.
Capy memory and Context7 docs lookup connect as external services. The author states the project was made with Claude Code but is reviewed and reworked by a human. GitHub does not recognize the license, so check the terms in `LICENSE.md`.
How to disable. For the plugin run `claude plugin uninstall kk@claude-toolbox`. For a full install remove the added `.claude/`, `.codex/` and `.mcp.json` files from the project and restore your own settings from git.
FAQ
How do I install only the skills?
Install the `kk` plugin with `claude plugin install kk@claude-toolbox`. The repository template is not needed.
Which skills form the pipeline?
`/kk:design`, `/kk:review-design`, `/kk:implement`, `/kk:review-code`, `/kk:test` and `/kk:document`, plus helpers `/kk:brainstorm` and `/kk:clarify-docs`.
Which languages do the skills cover?
Separate instructions exist for Go, Java, JS and TS, Kotlin, Kubernetes and Python.
Does it work with Codex?
Yes, the docs describe Claude Code and Codex support and the repository has `.codex/config.toml` and `hooks.json`.
What are Capy and Context7 for?
Capy keeps findings, decisions and conventions between sessions and Context7 pulls current library documentation.
Discussion
No questions or reviews yet.
Related
PluginsNick Tune's plugin marketplace for Claude Code: checks before a task is handed in, code review and TDD personas
Claude Code Blueprint: 12 agents, 19 skills, 16 hooks and 6 CLAUDE.md rules, copy what you need or install a preset
Claude Codex Settings
claude-codex-settings
A marketplace of dozens of plugins for Claude Code, Codex, Cursor and Gemini: skills, commands, hooks, subagents and configs for many tasks
The ce plugin for Claude Code and Codex: eleven skills and four sub-agents for plans, review, tests, architecture and writing.
A plugin that readies a project for Claude Code: guidance, tests, review, TDD, refactoring and a resumable fix loop
A set of 116 skills, 44 agents and safety hooks for Claude Code, Cursor, Codex, Gemini CLI and nine more editors
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| claude-toolbox (this card) | no | 155 | |
| claude-config | no | 916 | a set of Claude Code rules and configs, with no skill pipeline with review |
| superpowers-zh | no | 8,534 | also a skill pipeline for development but without a repository template or settings sync |
README in short
The repository combines a configuration template, the `kk` plugin, documentation and examples of real sessions. The author stresses that skills are reviewed by a human and gives step-by-step guides for a new project, an existing repository and a plugin-only install.
README badge
Original SKILL.md
--- name: brainstorm description: | TRIGGER when: asked to brainstorm, think through, explore, or pressure-test a technical idea or decision through conversation. Interviews adaptively, asks one question at a time, and closes with a chat recap. For technical products, architecture, infrastructure, tools, operations, and engineering workflows. Written design/specification and task-list requests belong to /kk:design; durable domain glossary/reference kits to /kk:model; implementation and fixes to /kk:implement. Not for nontechnical brainstorming. --- # Brainstorm a Technical Idea Help the user reach their next decision: pursue, discard, narrow, or investigate an idea. Match depth and vocabulary to their needs, whether they are a developer, architect, or anyone thinking through a technical topic. An implementation-ready specification is not required. Keep the working context and result in the conversation. Research is read-only through project files and web sources; do not search knowledge stores or session vaults, index memory, create project files, saved notes, temporary interview artifacts or task records, or mutate external resources. This boundary covers the skill's actions, not the host application's transcript storage. ## Workflow **Mandatory order — instructions before engagement.** The entry steps are strictly sequential: fully load this SKILL.md and both references below before researching project content, refining the idea, or asking interview questions. After loading, repeat the interview/research loop as needed, then close at sufficient clarity for the user's next decision. 1. **Load reasoning references.** Read [shared-ideation-frameworks.md](shared-ideation-frameworks.md) and [shared-idea-refinement-criteria.md](shared-idea-refinement-criteria.md). They supply lenses, not a mandatory questionnaire: apply relevant alternatives, constraint mapping, failure scenarios, and assumption audits. Product-market, adoption, differentiation, and MVP questions are context-dependent; omit them when they do not help the technical decision. 2. **Establish the current decision.** Use the idea and context already supplied. Ask a framing question only to resolve a real ambiguity about the problem or next decision; no ceremonial confirmation, persona interview, or numerical success metric is required. If the requested output is ambiguous between conversation and a written artifact, clarify that output before assuming permission to create anything. 3. *