Catalog¶
This is the shopping list. Skills, Plugins, MCP & LSP explains what these things are; this page is what to install and reach for. It covers how plugins arrive from a marketplace, every plugin this team actually runs — grouped by the job it does — and a quick router for picking the right one under pressure.
Who this is for. Engineers setting up their machine or deciding which tool fits the task in front of them. If a term here is unfamiliar (skill, MCP, LSP, subagent), the previous page defines it.
Installing plugins from a marketplace¶
A marketplace is a catalog of plugins (often a GitHub repo). The flow:
# Add a marketplace (Anthropic's official one is registered automatically on first launch)
claude plugin marketplace add anthropics/claude-plugins-official
# Browse and install interactively
/plugin
# …or install a specific plugin
/plugin install superpowers@claude-plugins-official
After installing, run /reload-plugins to pick up skills, agents, hooks, and servers without restarting. To develop or test a local plugin, launch with claude --plugin-dir ./my-plugin. This team also has a couple of non-default marketplaces registered (the OpenAI Codex marketplace openai/codex-plugin-cc, and supermemoryai/claude-supermemory).
This team's installed plugins¶
Every plugin below is actually installed and enabled in this setup. Use the "reach for it when" column as a quick router.
Workflow¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
superpowers (claude-plugins-official) |
A suite of workflow skills: brainstorming, writing-plans, executing-plans, subagent-driven-development, test-driven-development, systematic-debugging, using-git-worktrees, requesting-code-review, receiving-code-review, verification-before-completion, dispatching-parallel-agents, finishing-a-development-branch, writing-skills |
You want a disciplined, reviewable process instead of vibe-coding. Deep-dived in Superpowers and Workflows. |
ralph-loop (claude-plugins-official) |
Runs a prompt or slash command repeatedly on a loop (/ralph-loop, /cancel-ralph) |
A task needs the same prompt applied iteratively until a condition is met (polling, repeated passes). |
skill-creator (claude-plugins-official) |
Create, edit, optimize, and eval custom skills | You've repeated a workflow enough times to package it. See Reusable Skills and Automation. |
Review and quality¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
code-review (claude-plugins-official) |
/code-review — reviews a PR or your current working diff for correctness and quality |
Before merging, or to get a second read on a change you just made. |
codex (openai-codex) |
Delegates a task to OpenAI's Codex/GPT through a shared runtime as a second opinion or rescue pass (/codex:rescue, /codex:setup, codex-rescue agent) |
Claude is stuck, or you want an independent model to diagnose, implement, or adversarially review. |
code-simplifier (claude-plugins-official) |
Simplifies recently changed code for clarity and consistency without changing behavior | After a feature lands and you want to tidy it — quality only, not bug hunting. |
claude-md-management (claude-plugins-official) |
Audits and improves CLAUDE.md files (/revise-claude-md, claude-md-improver) |
Your project memory has drifted, grown stale, or needs a quality pass. |
codexhere is the plugin, not the standalone CLI. This entry runs Codex inside Claude Code as a delegated helper. Codex also ships as its own standalone terminal agent you'd run instead of Claude Code — covered, alongside OpenCode, in Codex and OpenCode.
Docs and knowledge¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
context7 (claude-plugins-official) |
MCP server that fetches current docs for a library/framework/SDK by name | You're using a library where the model's training data may be out of date. Prefer this over guessing API syntax. |
Security and audit¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
entry-point-analyzer (trailofbits) |
Finds state-changing entry points in smart contracts and categorizes them by access level | Auditing Solidity/Vyper/Solana/Move/CosmWasm contracts; mapping external functions and privileged operations. |
ask-questions-if-underspecified (trailofbits) |
Forces clarifying questions before implementing when requirements are ambiguous | A task is underspecified and guessing would be expensive. |
From the Trail of Bits marketplace. Both plugins above come from Trail of Bits' security-focused skills marketplace —
trailofbits/skills— not the official catalog. It ships more than these two; if you do security or smart-contract work, browse the whole set (see also Further Reading).Beyond this team's set. The official catalog and Trail of Bits aren't the whole world of skills. Practitioners publish their own —
mattpocock/skillsis a strong, model-agnostic collection for everyday engineering (aligning with the agent, domain modeling, TDD, review), andeveryinc/compound-engineering-pluginpackages the compounding-workflow idea from Reusable Skills and Automation. Browse them for patterns to adapt, not just to install wholesale.
Language servers (LSP)¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
typescript-lsp (claude-plugins-official) |
Real TypeScript language-server diagnostics | Working in TS/JS and you want type errors caught precisely, not approximated. |
pyright-lsp (claude-plugins-official) |
Pyright diagnostics for Python | Python projects — type checking and symbol resolution. |
rust-analyzer-lsp (claude-plugins-official) |
rust-analyzer diagnostics | Rust projects. |
Why LSP plugins matter. They give Claude the same real-time diagnostics your IDE shows — go-to-definition, type errors, unused symbols — instead of inferring them from text. Fewer hallucinated APIs, faster correct fixes.
App-building¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
agent-sdk-dev (claude-plugins-official) |
Scaffolds and verifies Claude Agent SDK apps (/new-sdk-app, plus TS/Python verifier agents) |
Building your own agent on top of Claude Code's engine. Pairs with Integrating AI via API. |
frontend-design (claude-plugins-official) |
Guidance for distinctive, intentional UI design | Building or reshaping UI and you want it to not look templated. |
Misc / commerce¶
| Plugin | What it provides | Reach for it when |
|---|---|---|
stripe (claude-plugins-official) |
Stripe integration guidance, an MCP server, and skills (test-cards, explain-error, stripe-best-practices, Connect helpers) |
Building or reviewing any Stripe payments integration. |
Want X → use Y (quick router)¶
- Plan a non-trivial feature →
superpowers:brainstormingthenwriting-plans. - Get a second opinion / rescue a stuck task →
codex. - Review before merge →
/code-review. - Use a fast-moving library correctly →
context7. - Tidy code you just wrote →
code-simplifier. - Audit a smart contract →
entry-point-analyzer. - Catch type errors precisely → the matching LSP plugin.
- Package a repeated workflow →
skill-creator.
Related¶
- Skills, Plugins, MCP & LSP — what these things are, if you skipped ahead
- Superpowers and Workflows · MCP and Config
- Onboarding a New Member — installing the team's toolset on day one
Sources¶
- Discover and install plugins — https://code.claude.com/docs/en/discover-plugins
- Create plugins — https://code.claude.com/docs/en/plugins
- Commands reference — https://code.claude.com/docs/en/commands
- Trail of Bits — Claude skills marketplace — https://github.com/trailofbits/skills