REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
1Password for coding agents: find, use, and save secrets without exposing them
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/kitlangton-2password)在同一工作中把 AI 连到服务的 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
同类工作,排序与 Skill 列表相同
| 本项 2password | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | kitlangton | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 329 | 465,460 | 233,910 | 149,364 | 144,149 |
| 最近代码更新 | 2026年10月4日 | 2026年11月17日 | 2026年4月20日 | 2026年11月17日 | 2026年11月17日 |
| 许可证 | MIT | MIT | – | Apache-2.0 | MIT |
kitlangton 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
开发者未说明 · SKILL.md 形式的 Skill 可用于支持 Agent Skills 的应用,如 Claude Code、Codex 和 Cursor
GitHub account kitlangton (person), 12 years old
★ 329 stars, few users so far · code last changed 2026年10月4日
No, plain text files the AI reads and follows
No scripts, so it does not go online by itself
No hooks or writes to machine settings found
Checked automatically from the repo files at commit e011177 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Uploading to claude.ai would fail: the description is over 200 characters (326). Use "Let the AI install it" instead
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /2password,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
把 2password 文件夹放到 ~/.agents/skills/2password
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 2password
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
6,807 个字符 · 在 GitHub 打开完整文件
---
name: 2password
description: Use for any password, API key, token, credential, secret, or 1Password task, including finding credentials, injecting them into commands or env files, saving new API keys, checking or updating login passwords, auditing vaults, and setting up unattended access. Prefer the 2password CLI over raw op whenever it covers the task.
---
# 2password
`2password` wraps the 1Password CLI (`op`) so you can work with secrets without seeing them, and with as few 1Password prompts as possible. All output is JSON.
## Rules
- Never print, log, summarize, or repeat secret values. Pass `op://` references around instead.
- Any 1Password call may show the user an approval prompt, and each new shell you open can need a fresh approval. Batch a task's work into as few commands as possible: many queries in one `find`, many references in one env file run with `env run`.
- Never run `op whoami` or any other preflight check. Just run the command; the desktop app authorizes it when needed.
- Never retry a write that failed or reported "unverified". Run `find` first to check whether it already happened.
## Find credentials
```bash
2password find openai anthropic "github actions" # many queries, one prompt
2password find stripe --vault Work --account my.1password.com
```
Each match appears once as `{ ref, title, kind }`, with `queries` when you asked several. A query that matches nothing returns `suggestions` with close titles (typos included) from the same lookup. Use those rather than searching again or listing whole vaults. It never returns values.
## Use credentials
```bash
2password run --env "OPENAI_API_KEY=op://Personal/OpenAI API Key/credential" -- bun run dev
2password env write .env.tpl "OPENAI_API_KEY=op://Personal/OpenAI API Key/credential" "STRIPE_KEY=op://Work/Stripe API Key/credential"
2password env run .env.tpl -- bun run dev # preferred: no plaintext on disk
2password env resolve .env.tpl --output .env # only when a real file is required (mode 0600)
2password read "op://Personal/OpenAI API Key/credential" # last resort: prints the value
```
All references in a template resolve with one prompt. Never loop over `read`. A `.env.tpl` that holds only `op://` references is safe to inspect; a resolved `.env` is plaintext.
`run` and `env run` keep the secret out of argv, the template, and 2password's own output, but the selected child process receives plaintext in its environment. Treat that child as a trusted secret consumer: do not inject credentials into environment-dump/debug commands or helpers whose purpose is to reveal the value.
## Save a new API key
```bash
2password create api-credential --title "OpenAI API Key" --vault Personal --clipboard
some-command-that-prints-a-key | 2password create api-credential --title "OpenAI API Key" --vault Personal --stdin
```
- `--vault` is required. Use the vault the user names, or their documented default; ask if neither exists.
- Never put the value in an argument, a command substitution, or a temporary file, and never read the clipboard into the conversation first.
- Optional `--url`, `--notes`, and `--account` take non-secret metadata only.
- Success means exit 0 and `"verified": true`. Reuse the returned `ref`; don't `read` it to double-check.
- Duplicate titles in the vault are refused, and nothing is ever overwritten.
## Check or update a login password
```bash
2password password "Example Airline" --vault Personal --clipboard # compare only
2password password "Example Airline" --vault Personal --clipboard --apply # update, then verify
```
The command refuses logins that have passkeys or unnamed imported fields. Use `--repair-imported-fields` only after the user approves it.
## Audit a vault
```bash
2password inventory --vault Work # item and field metadata, never values or URL query strings
2password audit --vault Personal # duplicate titles, untagged machine credentials, old logins, transient URLs
```
Treat findings as candidates for review. Propose renames or changes, and only make them after the user approves.
## No prompts: service account (macOS, Windows)
If the user is tired of approval prompts, suggest a service account. Only set one up when the user asks.
```bash
2password service-account setup --vault Automation --create-vault --write --save-vault Personal
```
After setup, every command authenticates silently with a token stored in macOS Keychain (on Windows, a DPAPI-encrypted file under `%LOCALAPPDATA%`), but it can only reach the Automation vault. On Windows, any process running as the user can decrypt that file, so the vault's narrow scope is the real boundary. Keep the credentials agents use in that vault.
- `2password --desktop <command>` uses the normal desktop login, for example to reach other vaults.
- `service-account status | connect --clipboard | recover | forget`. `forget` removes only this Mac's copy; it doesn't revoke the account.
- If setup fails partway, don't rerun it. Run `2password service-account status`.
- `OP_SERVICE_ACCOUNT_TOKEN` in the environment overrides the saved account (Linux, CI).
- `run` and `env run` remove the service-account token from the child process's environment.
## Item conventions
- API keys use the API Credential category, the built-in `credential` field, and a title like `<Provider> API Key` or `<Provider> <Purpose> API Key`.
- Logins use the built-in `password` field and a title like `<Provider>` or `<Provider> <Account>`.
- The vault shows who owns the item, so don't repeat the vault name in the title. Keep one current secret per item.
## Report anything that goes poorly
2password is built for agents, so your experience is how it improves. Speak up if a command fails unexpectedly or you had to work around it. The same goes for more approval prompts than expected, output that's awkward or too verbose, or anything in this skill that's unclear or wrong. Tell the user, and offer to open an issue. It posts publicly from their GitHub account, so get their okay first.
```bash
gh issue list --repo kitlangton/2password --state all --search "<keywords>" # add to an existing issue instead of duplicating it
gh issue create --repo kitlangton/2password --title "<what went wrong>" --body "<details>"
```
Include the output of `2password doctor` (versions and setup; it never prompts and is safe to share), the command you ran, and what you expected versus what happened. Never include secret values, item titles, vault or account names, or `op://` references. Replace them with placeholders.
## Everything else
Use raw `op` for other item categories, editing, moving, sharing, deleting, and vault management. Pass plaintext through JSON templates on stdin, never in arguments. After discovery, refer to items and vaults by ID.
提问、分享技巧或报告无法使用。按“有帮助”数量排序。