Context7
Up-to-date docs for AI agents
作者提供的说明(英文原文)
Shared team context for ProductNow: search, read, write and organise workspace documents.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"teamctx": {
"command": "npx",
"args": [
"-y",
"@reneturcios/teamctx",
"mcp"
]
}
}
}在终端中运行这条命令
claude mcp add teamctx -- npx -y @reneturcios/teamctx mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"teamctx": {
"command": "npx",
"args": [
"-y",
"@reneturcios/teamctx",
"mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"teamctx": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@reneturcios/teamctx",
"mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月6日
注册名称已通过 GitHub 账户 CloudCorpRecords 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(0.1.0)于4天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js, Docker。
同一分类中有:Context7, Codebase Memory, Exa
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Up-to-date docs for AI agents
Codebase knowledge graph for AI agents — 162 languages, sub-ms queries, 99% fewer tokens.
Connect AI agents to Exa for web search, content fetching, and multi-step research.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
teamctx is a forwarding client for ProductNow's shared team context. The real ProductNow server is hosted, not self-hosted: teamctx takes MCP tool calls (or CLI calls) for searching, reading, writing, and organising workspace documents and forwards them to the hosted ProductNow MCP endpoint, returning whatever that server returns. One binary serves a command-line interface and a stdio MCP server.
The upstream endpoint is https://api.productnow-prod.com/mcp (Streamable HTTP), protected by OAuth 2.0 via Auth0 with the required scope mcp:use. teamctx does not run the OAuth flow itself: you supply an access token you already hold through the PRODUCTNOW_ACCESS_TOKEN environment variable, and teamctx sends it as a Bearer credential on every forwarded call. If the variable is missing or blank, teamctx fails with an error naming it before any request is attempted.
This package implements 16 of the 18 enumerated upstream tools — the labelled core subset for which the public docs publish a full Input table. Two tools are deliberate omissions because no complete parameter schema is published for them: curate_knowledge_pack and create_help_document. They are not advertised and not forwarded. Parameter types for the 16 built tools are inferred from the Input tables in the public documentation (the upstream tool catalogue publishes no JSON-Schema property definitions); names, requiredness, and enum values are as published.
npx @reneturcios/teamctx call search '{"query":"pricing"}'
Or run the container image:
docker run -i ghcr.io/cloudcorprecords/teamctx
The container defaults to the MCP server (see below). Pass CLI arguments to use it as a CLI instead.
The CLI uses one generic subcommand rather than a subcommand per tool, because every tool takes a JSON-object argument set:
teamctx call <tool> [json-args]
teamctx call search '{"query":"pricing","resultType":"documents"}' teamctx call fetch '{"documentId":"doc_123"}' teamctx call fetch_folder '{}' teamctx call remember '{"content":"We ship on Fridays."}'
teamctx --version
json-args must be a single JSON object when given; invalid JSON, a non-object value, an unknown tool, or extra arguments are usage errors (exit 2) and are rejected before any credential check. Runtime failures — including a missing PRODUCTNOW_ACCESS_TOKEN or an upstream error — exit 1. Success exits 0 and prints the upstream result as JSON.
One bin serves both faces. The mcp subcommand starts a stdio MCP server:
npx -y @reneturcios/teamctx mcp
MCP client configuration:
{
"mcpServers": { "teamctx": { "command": "npx", "args": ["-y", "@reneturcios/teamctx", "mcp"], "env": { "PRODUCTNOW_ACCESS_TOKEN": " " } } } }
| Tool | Required inputs | What it does |
|---|---|---|
search | — (query required for folder search) | Search shared context or ProductNow help for evidence excerpts with citations, or search folder names. |
fetch | documentId | Read one document: full text, sections, comment threads, folder, canonical URL. |
fetch_folder | — | List a folder or the warehouse root, optionally as a nested tree (deep). |
list_document_versions | documentId | List every version of a document, newest first; optional version-status filter. |
get_status | documentId | Report whether a document's draft agent is running or idle. |
remember | content | Store one fact, decision, or note in the right document/folder. |
create_document | name, prompt | Create a document, file it in a folder, and generate content with AI. |
edit_document | documentId, message | Hand editing instructions to the document's own editing agent. |
update_document_status | documentId, status | Promote the current draft with an action verb: snapshot, review, or publish. |
create_folder | name | Create a folder, optionally under a parent folder. |
move | ids | Move documents and folders, several at once, into one folder or the root. |
rename | id, name | Change one document's or folder's name. |
archive | ids | Archive documents and folders, several at once; nothing is deleted. |
comment_on_document | documentId, comment | Comment on a document's current review version, optionally anchored to quoted text. |
reply_to_thread | threadId, message | Reply to an existing comment thread. |
upload_media | fileName, mimeType | Allocate a signed upload URL (valid 5 minutes) for image/video media. |
Notes on the less obvious rules, as documented upstream: search defaults to the workspace scope and documents result type; creator and date filters combine as documented in the tool schema. list_document_versions filters by *version* statuses (DRAFT … ARCHIVED), while update_document_status takes *action verbs* (snapshot / review / publish) — do not mix them up. create_document's folderId is required in practice: omit it only for the warehouse root.
npm install
npm run build # tsc -> dist/ npm test # node test runner over test/*.test.ts, no network (fake fetch only) npm run dev # run the CLI from source via tsx
MIT — see LICENSE.