T
teamnebula-ai/tmn-fdekit
Team Nebula's forward-deployed engineering kit: staff-level skills, guard hooks and zero-token tracing for…
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": {
"teamnebula-ai-tmn-fdekit": {
"command": "npx",
"args": ["-y", "github"]
}
}
}解析自我们于 2026-11-09 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": {
"teamnebula-ai-tmn-fdekit": {
"command": "npx",
"args": ["-y", "github"]
}
}
}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 teamnebula-ai-tmn-fdekit -- npx -y github
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
- 0 ★ · teamnebula-ai/tmn-fdekit · 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 2026-11-10 · 最近见到 2026-11-14 · 共见到 5 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 1 个来源 · repo
- 安装
npx -y github
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 -y github。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
teamnebula-ai/tmn-fdekit；我们收录于 2026-11-10；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。# tmn-fdekit
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
Staff-level engineering discipline for coding agents, from Team Nebula.
Agents don't usually fail by writing bad code. They fail by saying something
untrue with confidence: "tests pass", "merged", "reviewed". They also approve
their own work, post ten comments where one would do, and call a diff-only
read a review. This plugin makes the good habits the easy path, and makes the
amateur ones impossible.
Skills hold the judgement: how to review adversarially, prove a change
works, ship a small honest PR, answer review, debug from evidence, harden,
write for people, and design UI that doesn't look generated.
Hooks and scripts hold every rule that is a yes/no on data, so no model
has to remember it and no prompt can talk its way past it.
Tracing records everything the agent did at zero model-token cost, and
can send Claude Code's own telemetry to Langfuse, Phoenix or any
OpenTelemetry backend.
> Use at your own judgement. Agents with these skills push code, post
> reviews and send messages. You're responsible for what they do. Read
> DISCLAIMER.md first.
## Install (for people)
You need git, python3 3.10+ and, for the installer, node 18+. gh is
needed for the GitHub checks, and docker only for compose projects.
Claude Code gets everything: skills, review agents and every hook.
bash
claude plugin marketplace add teamnebula-ai/tmn-fdekit
claude plugin install tmn-fdekit@tmn-fdekit
Restart Claude Code, then type /tmn-fdekit: to see the skills. To update
later: claude plugin marketplace update tmn-fdekit && claude plugin update tmn-fdekit@tmn-fdekit.
Codex, Cursor, Gemini CLI, Copilot, Windsurf/Devin, OpenCode or Amp:
bash
npx -y github:teamnebula-ai/tmn-fdekit install # finds the agents on this machine
npx -y github:teamnebula-ai/tmn-fdekit install --target codex,cursor --project # into this repo only
npx -y github:teamnebula-ai/tmn-fdekit hooks --target codex # Codex runs the guard hooks too
Claude.ai or Claude Desktop: run npx -y github:teamnebula-ai/tmn-fdekit zip,
then upload each zip in Customize › Skills.
Check it worked: npx -y github:teamnebula-ai/tmn-fdekit doctor lists the
tools it found and where the skills are installed.
Then fit it to your team: ask your agent to "set up the kit". See
Make it yours. After that, ask for real work ("review PR
#12", "make docs for this repo") and it picks the matching skill.
To pin a version, add the tag: github:teamnebula-ai/tmn-fdekit#v0.6.1. To
remove it: npx -y github:teamnebula-ai/tmn-fdekit uninstall (and
hooks --remove), which only deletes what it installed.
## Install (for agents)
Agents: follow INSTALL.md. It works the same in Claude Code,
Codex, Antigravity, Cursor, OpenCode, Gemini CLI, Copilot, Windsurf/Devin, Amp
and any agent that reads ~/.agents/skills: check the tools, pick the target,
install, confirm with doctor, then offer tmn-setup. You can also point an
agent at this repo and say "install this kit"; AGENTS.md, .claude/CLAUDE.md,
GEMINI.md and llms.txt all send it to the same steps.
Once installed, load tmn-staff-bar at the start of real engineering work.
It sets the ground rules, the run mode and the team rules the other skills
follow.
| Doc | Covers |
| Getting started | install to first receipt in ten minutes |
Agents don't usually fail by writing bad code. They fail by saying something
untrue with confidence: "tests pass", "merged", "reviewed". They also approve
their own work, post ten comments where one would do, and call a diff-only
read a review. This plugin makes the good habits the easy path, and makes the
amateur ones impossible.
Skills hold the judgement: how to review adversarially, prove a change
works, ship a small honest PR, answer review, debug from evidence, harden,
write for people, and design UI that doesn't look generated.
Hooks and scripts hold every rule that is a yes/no on data, so no model
has to remember it and no prompt can talk its way past it.
Tracing records everything the agent did at zero model-token cost, and
can send Claude Code's own telemetry to Langfuse, Phoenix or any
OpenTelemetry backend.
> Use at your own judgement. Agents with these skills push code, post
> reviews and send messages. You're responsible for what they do. Read
> DISCLAIMER.md first.
## Install (for people)
You need git, python3 3.10+ and, for the installer, node 18+. gh is
needed for the GitHub checks, and docker only for compose projects.
Claude Code gets everything: skills, review agents and every hook.
bash
claude plugin marketplace add teamnebula-ai/tmn-fdekit
claude plugin install tmn-fdekit@tmn-fdekit
Restart Claude Code, then type /tmn-fdekit: to see the skills. To update
later: claude plugin marketplace update tmn-fdekit && claude plugin update tmn-fdekit@tmn-fdekit.
Codex, Cursor, Gemini CLI, Copilot, Windsurf/Devin, OpenCode or Amp:
bash
npx -y github:teamnebula-ai/tmn-fdekit install # finds the agents on this machine
npx -y github:teamnebula-ai/tmn-fdekit install --target codex,cursor --project # into this repo only
npx -y github:teamnebula-ai/tmn-fdekit hooks --target codex # Codex runs the guard hooks too
Claude.ai or Claude Desktop: run npx -y github:teamnebula-ai/tmn-fdekit zip,
then upload each zip in Customize › Skills.
Check it worked: npx -y github:teamnebula-ai/tmn-fdekit doctor lists the
tools it found and where the skills are installed.
Then fit it to your team: ask your agent to "set up the kit". See
Make it yours. After that, ask for real work ("review PR
#12", "make docs for this repo") and it picks the matching skill.
To pin a version, add the tag: github:teamnebula-ai/tmn-fdekit#v0.6.1. To
remove it: npx -y github:teamnebula-ai/tmn-fdekit uninstall (and
hooks --remove), which only deletes what it installed.
## Install (for agents)
Agents: follow INSTALL.md. It works the same in Claude Code,
Codex, Antigravity, Cursor, OpenCode, Gemini CLI, Copilot, Windsurf/Devin, Amp
and any agent that reads ~/.agents/skills: check the tools, pick the target,
install, confirm with doctor, then offer tmn-setup. You can also point an
agent at this repo and say "install this kit"; AGENTS.md, .claude/CLAUDE.md,
GEMINI.md and llms.txt all send it to the same steps.
Once installed, load tmn-staff-bar at the start of real engineering work.
It sets the ground rules, the run mode and the team rules the other skills
follow.
| Doc | Covers |
| Getting started | install to first receipt in ten minutes |
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/teamnebula-ai-tmn-fdekit/)
- 来源
- github · https://github.com/teamnebula-ai/tmn-fdekit
- 按领域浏览
- Agent 技能
- 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。