Claude Kit plugin
Evidence discipline for Claude Code: a claim without an artifact is rejected and tests are re-run before anything is called done.
- 97 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Hooks and agents run commands and edit project files, and `init` creates configuration with MCP servers
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/duthaho-claudekitDetects 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.
- Hooks and agents run commands and edit project files, and `init` creates configuration with MCP servers
This is third-party code. Review the repository files before installing.
What it does
Claude Kit is a Claude Code plugin with 15 skills, 8 agents and 5 output styles. The skills form a five-phase chain: investigate, design, implement, verify, ship. Each phase must produce an artifact with evidence, such as file:line citations or the output of executed tests. At its core is plan review by two parallel reviewers, one for architecture and one for experience, and the `verification-gate` skill runs before any claim of done. The `/claudekit:init` command sets up rules, hooks (formatting, dangerous command blocking, secret detection, sensitive file guard) and MCP configs in the project.
Who it is for. Senior engineers and tech leads who want verification discipline in Claude Code without extra ceremony.
Good fit when
- A new feature: a path from specification through a plan to verified code.
- A bug with no known cause: an investigation with file:line citations.
- A refactor where preserved behaviour is proven with test measurements.
- A library release: a dependency audit and a changelog built from the real diff.
Not a fit when
- A one-line change where even the shortened chain feels excessive.
- The agent is not Claude Code: the plugin is built for it.
- A quick experiment with no plans or review.
Example request
Run `/claudekit:shape-spec` for a report export feature, write a plan and run plan review before any code.Limitations
Built for Claude Code 1.0 and newer, with Git and Node.js or Python. The `/claudekit:init` wizard creates rules, hooks and a `.mcp.json` with Context7, Sequential, Playwright, Memory and Filesystem in the project, so review the changes in `.claude/`. Hooks block some commands and can occasionally get in the way of legitimate actions. The chain can be shortened by task size, but the README advises against dropping evidence checks.
How to disable. Remove the plugin from the `/plugin` menu. Delete the rules, hooks and `.mcp.json` files created by `/claudekit:init` from the project by hand.
FAQ
What phases make up the chain?
Five: investigate, design, implement, verify and ship, plus a one-time `init` setup. Each phase requires a concrete artifact.
What does verification-gate do?
It runs the tests, checks the negative path, exercises the change outside the IDE and cross-checks the result with the original ask. Without it, done does not count.
What does /claudekit:init create?
Rules in `.claude/rules/`, hooks in `.claude/hooks/` and `settings.local.json`, plus a `.mcp.json` with five MCP servers. Output styles load on their own and need no such step.
How do I shorten the chain for a small change?
By change size: trivial edits skip `shape-spec`, `write-plan` and `plan-review`. `verification-gate` and `code-review-loop` stay.
How do I validate the plugin itself?
Run `node scripts/validate-plugin.cjs`. The gate itself is tested by `node scripts/test-verify-evidence.cjs`.
Discussion
No questions or reviews yet.
Related
PluginsA ready-made Claude Code setup: agents, slash commands, skills, safety hooks, rules and MCP connections in one install
Three Claude Code plugins: the savvy-flow orchestrator with tiered subagents and two mods for a progress panel and a cache timer
The ce plugin for Claude Code and Codex: eleven skills and four sub-agents for plans, review, tests, architecture and writing.
A plugin that turns Claude Code into a 16 agent team: documentation first planning, TDD and 9 pre-merge quality gates.
ClaudeKit Skills
ClaudeKit - Agent Skills
A plugin marketplace for Claude Code with a set of thirty-plus skills: frontend, backend, devops, databases, documents, debugging and MCP work
Claude Code plugin with 17 skills: task sizing, design review, TDD, fresh-context subagents and a PR after three human decisions.
README in short
README sections follow common tasks: feature, bug fix, refactor, code exploration, dependency audit and release. Each chain has a clear stopping point.