L
lawzava/megapowers
One plugin for Claude Code and Codex: task skills, an output style, a destructive-command tripwire, and a…
在 GitHub 查看 ↗链接已于 2026-11-12 核查 · HTTP 200 · 打开的是厂商自己的官网
安装与配置
客户端配置(取自该仓库的安装命令)
{
"mcpServers": {
"lawzava-megapowers": {
"command": "npx",
"args": ["skills"]
}
}
}解析自我们于 2026-10-26 抓取的该仓库 README;我们未执行过。
command/args 来自仓库写明的安装命令;不确定时请以该项目 README 为准。
VS Code(.vscode/mcp.json —— 键名是 servers)
{
"servers": {
"lawzava-megapowers": {
"command": "npx",
"args": ["skills"]
}
}
}VS Code 顶层键是 <code>servers</code>,其余与上面那份相同。
Claude Code(命令行)
claude mcp add --transport stdio lawzava-megapowers -- npx skills
Claude Code 不收 JSON 文件,用这条命令直接加。
配置项的键名以你的客户端为准:客户端与 CLI
- ★ 数取值
- 8 ★ · lawzava/megapowers · 取自 GitHub 搜索 API · (2026-11-13 20:00:50 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-10-24 · 最近见到 2026-11-14 · 共见到 25 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 8 ★ · 1 个来源 · repo
- 安装
npx skills
常见问题(答案都来自我们的记录)
怎么安装?
该仓库给出的安装命令是
npx skills。我们的记录里它是仓库自己提供的;不保证在你的客户端上直接可用。它适合谁 / 用来做什么?
我们把它归在这些标签下: Agent 技能。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
lawzava/megapowers;我们收录于 2026-10-24;最近见到 2026-11-14。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。megapowers
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
Plan, test, debug, and ship with more care in Claude Code and Codex.
One plugin, 15 skills, and safety hooks. About 1.4k tokens per session.
Install ·
What changes ·
Why megapowers ·
Skills ·
Docs
Coding agents are fast, but they can skip steps: edit before they know the
cause, say "done" without re-running the tests, or push and deploy without a
second look. megapowers gives Claude Code and Codex a skill (a short
playbook) for each of those moments. When your request matches one, the agent
is reminded to load it and follow it. Before the first commit or deploy in a
session, a hook pauses once and names the skill to load first.
It is exactly one plugin for Claude Code and Codex. It uses each tool's own
agents, permissions, worktrees, and memory, and ships no daemon, scheduler,
model router, formatter, or status line.
## What changes when it is installed
| You ask | What the agent does |
| "Why is this test flaky?" | systematic-debugging: finds the root cause before editing anything |
| "Add rate limiting to the API" | test-first-implementation: writes a failing test, sees it fail, then makes it pass |
| "Commit this and open a PR" | The first commit in the session pauses once and points the agent to verify-and-finish, which has it re-run your checks and report what actually passed |
| "Deploy to production" | safe-effects: confirms the exact target with you, then reads the result back |
| "Get a second opinion on this auth change" | independent-review: shows you what it will send, then asks a model from a different provider to review it |
| Agent runs rm -rf ~ or mkfs.ext4 /dev/sda1 | Blocked before it runs |
mermaid
flowchart TD
ask([You ask for something]) --> skill[Matching skill loads: plan, test first, debug, ...]
skill --> work[Agent works]
work --> gate{First commit, push, or deploy this session?}
gate -- yes --> check[Hook pauses once and names verify-and-finish or safe-effects]
check --> retry([Agent retries; command runs])
gate -- no --> run([Command runs])
work -. catastrophic command .-> blocked([Blocked])
Skills guide the agent; they do not force it (how often they load).
Only the pause and the command block are enforced, by the hook. Your
repository's own instructions and tools always come first.
## Install
Claude Code
bash
claude plugin marketplace add lawzava/megapowers@release
claude plugin install megapowers@megapowers
Codex
bash
codex plugin marketplace add lawzava/megapowers --ref release
codex plugin add megapowers@megapowers
Then start a fresh session. You need Go 1.25 or newer on PATH; the hooks
compile once into a local cache. Codex asks you to review and trust the hooks
before it runs them, and asks again whenever a release changes them.
Try it: ask "check my megapowers setup". The doctor reports the plugin
version, whether the hooks are registered, and whether the output style is
active.
Both installs track the release branch, which only moves to signed release
tags, so you never get unreleased main. Pinning, install scopes, and
uninstall are in docs/install.md.
## Why megapowers
Same behavior on Claude Code and Codex. One plugin, the same skills, and
the same Go hook code on both.
Checks before you ship. A hook pauses the first commit, push, PR, or
One plugin, 15 skills, and safety hooks. About 1.4k tokens per session.
Install ·
What changes ·
Why megapowers ·
Skills ·
Docs
Coding agents are fast, but they can skip steps: edit before they know the
cause, say "done" without re-running the tests, or push and deploy without a
second look. megapowers gives Claude Code and Codex a skill (a short
playbook) for each of those moments. When your request matches one, the agent
is reminded to load it and follow it. Before the first commit or deploy in a
session, a hook pauses once and names the skill to load first.
It is exactly one plugin for Claude Code and Codex. It uses each tool's own
agents, permissions, worktrees, and memory, and ships no daemon, scheduler,
model router, formatter, or status line.
## What changes when it is installed
| You ask | What the agent does |
| "Why is this test flaky?" | systematic-debugging: finds the root cause before editing anything |
| "Add rate limiting to the API" | test-first-implementation: writes a failing test, sees it fail, then makes it pass |
| "Commit this and open a PR" | The first commit in the session pauses once and points the agent to verify-and-finish, which has it re-run your checks and report what actually passed |
| "Deploy to production" | safe-effects: confirms the exact target with you, then reads the result back |
| "Get a second opinion on this auth change" | independent-review: shows you what it will send, then asks a model from a different provider to review it |
| Agent runs rm -rf ~ or mkfs.ext4 /dev/sda1 | Blocked before it runs |
mermaid
flowchart TD
ask([You ask for something]) --> skill[Matching skill loads: plan, test first, debug, ...]
skill --> work[Agent works]
work --> gate{First commit, push, or deploy this session?}
gate -- yes --> check[Hook pauses once and names verify-and-finish or safe-effects]
check --> retry([Agent retries; command runs])
gate -- no --> run([Command runs])
work -. catastrophic command .-> blocked([Blocked])
Skills guide the agent; they do not force it (how often they load).
Only the pause and the command block are enforced, by the hook. Your
repository's own instructions and tools always come first.
## Install
Claude Code
bash
claude plugin marketplace add lawzava/megapowers@release
claude plugin install megapowers@megapowers
Codex
bash
codex plugin marketplace add lawzava/megapowers --ref release
codex plugin add megapowers@megapowers
Then start a fresh session. You need Go 1.25 or newer on PATH; the hooks
compile once into a local cache. Codex asks you to review and trust the hooks
before it runs them, and asks again whenever a release changes them.
Try it: ask "check my megapowers setup". The doctor reports the plugin
version, whether the hooks are registered, and whether the output style is
active.
Both installs track the release branch, which only moves to signed release
tags, so you never get unreleased main. Pinning, install scopes, and
uninstall are in docs/install.md.
## Why megapowers
Same behavior on Claude Code and Codex. One plugin, the same skills, and
the same Go hook code on both.
Checks before you ship. A hook pauses the first commit, push, PR, or
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/lawzava-megapowers/)
- 来源
- github · https://github.com/lawzava/megapowers
- 按领域浏览
- Agent 技能
- Traction
- 8 ★
- 收录于
- 今天
同类推荐
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。