Team Kit

claude-code-team-kit

A port of cursor/plugins/cursor-team-kit to the Claude Code plugin format. Workflows for CI, code review, shipping, verification, and test reliability — no third-party service integrations required.

Original work © Cursor, MIT-licensed. This port © lukebward, also MIT-licensed. See NOTICE for details on what changed during the port.

Install

/plugin marketplace add lukebward/claude-code-team-kit
/plugin install team-kit@claude-code-team-kit

Skills

Skill	Description
loop-on-ci	Watch CI runs and iterate on failures until checks pass
review-and-ship	Run a structured review, commit changes, and open a PR
pr-review-canvas	Generate an interactive HTML PR walkthrough with annotated, categorized diffs
verify-this	Prove or disprove claims with baseline/treatment artifacts and a clear verdict
control-cli	Build or adapt a local harness to drive and profile interactive CLIs or TUIs
control-ui	Build or adapt a local browser/CDP harness for web or Electron UIs
make-pr-easy-to-review	Clean noisy PR history, improve descriptions, and add reviewer guidance
run-smoke-tests	Run Playwright smoke tests and triage failures
fix-ci	Find failing CI jobs, inspect logs, and apply focused fixes
new-branch-and-pr	Create a fresh branch, complete work, and open a pull request
get-pr-comments	Fetch and summarize review comments from the active pull request
check-compiler-errors	Run compile and type-check commands and report failures
what-did-i-get-done	Summarize authored commits over a given time period into a concise status update
weekly-review	Generate a weekly recap of shipped work with bugfix/tech-debt/net-new highlights
fix-merge-conflicts	Resolve merge conflicts, validate build/tests, and summarize decisions
deslop	Remove AI-generated code slop and clean up code style
workflow-from-chats	Extract durable working preferences from chats into skills, CLAUDE.md/AGENTS.md guidance, or workflow docs
thermo-nuclear-code-quality-review	Run an unusually strict maintainability review (code-judo, 1k-line rule, spaghetti, boundaries)

Agents

Agent	Description
ci-watcher	Monitor GitHub Actions runs and return concise pass/fail summaries (runs on haiku)
thermo-nuclear-code-quality-review	Subagent that runs the thermo-nuclear code quality rubric against a diff

What was not ported

The upstream Cursor kit ships two .mdc rule files (no-inline-imports, typescript-exhaustive-switch). Cursor rules are "always-applied" prompt fragments injected into context; Claude Code has no equivalent loaded-by-default rule system, so they were intentionally omitted. The closest Claude Code equivalents are project-level CLAUDE.md / AGENTS.md files, which you can add manually to your own project if you want similar guardrails.

A few Cursor-specific frontmatter fields (e.g. disable-model-invocation, is_background, model: fast) are left in place where the field is harmless, and translated where it isn't:

• model: fast → model: haiku
• is_background: true → removed (Claude Code controls background execution at invocation time)
• disable-model-invocation: true → left as-is (Claude Code ignores unknown frontmatter)

Attribution

All skill and agent content is the original work of the Cursor team. This repository only adapts packaging, frontmatter, and a small number of platform-specific references. The MIT license travels with the code; see LICENSE and NOTICE.