Agent-Native Calendar
Agent-Native Google Calendar - manage events, sync, and public booking
作者提供的说明(英文原文)
Search, read, upload and organize your private documents: receipts, contracts, tax records.
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 …”。
适用于网页版、桌面版和手机版 Claude:
https://api.pocketarchive.io/mcp在终端中运行这条命令
claude mcp add --transport http pocket-archive https://api.pocketarchive.io/mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"pocket-archive": {
"url": "https://api.pocketarchive.io/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": {
"pocket-archive": {
"type": "http",
"url": "https://api.pocketarchive.io/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。
ChatGPT 只能连接互联网上的服务器:
https://api.pocketarchive.io/mcp菜单名称可能因套餐和版本不同而略有差异。
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。
服务器要求先登录,才会列出它的工具。
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8日
注册名称已通过域名 pocketarchive.io 验证(注册中心会核实域名所有权)
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
在 254 毫秒内响应(HTTP 401)
声明支持 OAuth 登录,你的 AI 应用可以带你去登录
尚未检查;我们会在检查端点时读取工具列表
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 3 项适用的基础检查中通过了 3 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。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(1.0.0)于5天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
同一分类中有:Agent-Native Calendar, Excel MCP Server, MCP Server for Excel
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
Agent-Native Google Calendar - manage events, sync, and public booking
Create and edit Excel files with AI: charts, PivotTables, formulas. No Excel needed.
Excel automation for AI - Sheets, Power Query, DAX, VBA, Tables, Ranges and more. Windows only.
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。
Connect Claude, ChatGPT, Cursor, VS Code, Codex or any MCP app to your Pocket Archive, the AI document organizer for receipts, bills, contracts, warranties and tax documents. Your agent can search your documents, read their text, upload new files, track deadlines and organize records into Smart Collections, with read-only or read-and-write access you approve.
Server URL: https://api.pocketarchive.io/mcp · Streamable HTTP · OAuth sign-in · Setup guide on pocketarchive.io
This is a hosted (remote) MCP server; there is nothing to install or run locally. This repository contains its connection details, a reusable Skill and optional client adapters.
You need a Pocket Archive account. Create one for free. MCP access is included on every plan, including Free; uploads still count toward your plan's text recognition and storage limits.
Every app uses the same server URL: https://api.pocketarchive.io/mcp. The first time your app connects, it opens Pocket Archive in your browser. Check the app name, then choose read-only or allow changes.
claude mcp add --transport http pocket-archive https://api.pocketarchive.io/mcp
Then run /mcp in Claude Code, choose pocket-archive, and sign in.
Open Settings → Connectors and choose Add custom connector. Name it Pocket Archive, paste the server URL, and select Connect.
codex mcp add pocket-archive --url https://api.pocketarchive.io/mcp
codex mcp login pocket-archive
Add Pocket Archive to ~/.cursor/mcp.json:
{
"mcpServers": { "pocket-archive": { "url": "https://api.pocketarchive.io/mcp" } } }
Cursor opens your browser to sign in the first time it connects.
Add Pocket Archive to .vscode/mcp.json in your workspace, or run MCP: Open User Configuration from the Command Palette to add it for all workspaces:
{
"servers": { "pocket-archive": { "type": "http", "url": "https://api.pocketarchive.io/mcp" } } }
Start the server from the file or the Extensions view; VS Code opens your browser to sign in.
Turn on developer mode in Settings → Apps & Connectors → Advanced settings. Create a connector with the server URL and choose OAuth sign-in. Developer mode is a beta feature, and availability depends on your ChatGPT plan.
For other MCP apps, or to check compatibility:
| Setting | Value |
|---|---|
| Server name | pocket-archive |
| Endpoint | https://api.pocketarchive.io/mcp |
| Transport | Streamable HTTP |
| Authentication | Browser OAuth authorization code with PKCE S256 |
| Registration | Dynamic public-client registration (DCR), token_endpoint_auth_method=none |
| Resource | https://api.pocketarchive.io/mcp |
| Scopes | archive:read, optionally archive:write |
Use the client's connect/login action. It discovers OAuth metadata, registers its own name and callback URL, opens browser authorization, and exchanges the approved code for MCP credentials. The authorization page displays the registered client's name and callback host; each approved connection is granted and revoked independently.
If you are already signed in to Pocket Archive on the web in the same browser, the authorization page reuses that sign-in. Otherwise you sign in and return to the authorization request. The agent receives its own MCP tokens; it never sees your web session. Read-only connections expose read tools only. Settings → Connected apps in Pocket Archive lists and revokes connections.
A client needs remote Streamable HTTP and this OAuth flow to connect directly. A client supporting only stdio, legacy HTTP+SSE, confidential-client authentication or Client ID Metadata Documents needs an adapter. Not every agent product has been tested; compatibility is defined by these capabilities, not by product names.
Read-only connections see only the read tools. Allowing changes adds the write tools.
| Tool | Access | What it does |
|---|---|---|
search_documents | Read | Search by keyword with document type, tag, date, vendor and amount filters; returns snippets and citations |
list_documents_page | Read | List documents with filters, sorting and pagination |
list_documents | Read | List documents as a single array (legacy) |
get_document | Read | Get a document's fields, OCR text and processing state |
list_document_pages | Read | Read OCR text page by page, with page numbers for citations |
download_original_document | Read | Get short-lived download links for the original files |
get_document_upload_status | Read | Check an upload before and after it is finalized |
create_upload_session | Write | Reserve a signed upload URL for a PDF, JPEG, PNG or HEIC file up to 50 MiB |
create_document | Write | Finalize an upload and start OCR, analysis and indexing |
update_document | Write | Update a document's metadata or extracted fields |
delete_document | Write | Move a document to trash |
restore_document | Write | Restore a document from trash |
retry_document_processing | Write | Retry a failed processing stage |
| Tool | Access | What it does |
|---|---|---|
list_attention_items | Read | List Attention items (the Action Inbox) across the archive or for one document |
get_attention_item | Read | Get an Attention item and its related information |
create_attention_item | Write | Create an Attention item, optionally linked to a document |
update_attention_item | Write | Update an item's details, dates or status |
delete_attention_item | Write | Delete an Attention item and its reminders |
| Tool | Access | What it does |
|---|---|---|
list_collection_templates | Read | List collection templates and their setup questions |
list_collections | Read | List collections |
get_collection | Read | Get a collection, its readiness and details |
list_collection_documents | Read | List a collection's documents and suggestions |
list_collection_requirements | Read | List a collection's checklist |
list_collection_requirement_documents | Read | List the documents matched to a checklist item |
list_collection_attention_items | Read | List Attention items for a collection |
create_collection | Write | Create a collection from a template |
update_collection | Write | Update, archive or restore a collection |
delete_collection | Write | Delete a collection (its documents are kept) |
refresh_collection | Write | Re-evaluate a collection and its suggestions |
include_collection_document | Write | Add a document to a collection |
decide_collection_document | Write | Accept or exclude a suggested document |
remove_collection_document | Write | Remove a document from a collection |
create_collection_requirement | Write | Add a checklist item |
update_collection_requirement | Write | Update or complete a checklist item |
delete_collection_requirement | Write | Delete a checklist item |
confirm_collection_requirement_document | Write | Confirm that a document satisfies a checklist item |
decide_collection_requirement_document | Write | Accept or reject a suggested match for a checklist item |
remove_collection_requirement_document | Write | Clear a document from a checklist item |
skills/pocket-archive/SKILL.md contains agent-independent archive workflows: finding records, uploading files and waiting for processing, and downloading originals. An agent that supports SKILL.md can load this folder through its own skill installation mechanism. Other frameworks can use the instructions as task guidance alongside the discovered MCP tool schemas. MCP clients can call the tools without installing the Skill.
# Claude Code
mkdir -p ~/.claude/skills/pocket-archive && curl -fsSL https://raw.githubusercontent.com/ShisunXia/pocket-archive-mcp/main/skills/pocket-archive/SKILL.md -o ~/.claude/skills/pocket-archive/SKILL.md
mkdir -p ~/.codex/skills/pocket-archive && curl -fsSL https://raw.githubusercontent.com/ShisunXia/pocket-archive-mcp/main/skills/pocket-archive/SKILL.md -o ~/.codex/skills/pocket-archive/SKILL.md
In the Claude app, download pocket-archive.zip and upload it under Customize → Skills. Code execution must be turned on.
The Skill provides workflow guidance; the host supplies MCP discovery, OAuth, credential storage, tool calls and any binary upload/download capability. If the host cannot perform binary HTTP transfers, the agent directs the user to upload on the web rather than claiming that a local file path was uploaded.
mcp.json: Agent Plugins MCP configuration with a streamable-http server. Use it in hosts that support this format; other clients should configure the endpoint in their native format.skills/pocket-archive/SKILL.md: shared workflow instructions..codex-plugin/plugin.json and .mcp.json: optional Codex plugin adapter. The http type in this adapter maps to the same Streamable HTTP endpoint. Package this repository, including its dotfiles, as one plugin root following the Codex plugin format.skills/pocket-archive/agents/openai.yaml: optional OpenAI display and dependency metadata; other agents can ignore it.server.json: entry for the official MCP Registry.glama.json: maintainer claim for Glama.A configuration file is not installed into an agent automatically. Include the adapter and Skill your client needs when packaging for it. This repository contains no credentials.
Authentication follows the MCP authorization specification. Document text is your data and agents are instructed to treat it as data, never as instructions. See the Pocket Archive privacy policy for how your data is handled.
只显示了开头部分;其余内容请在仓库中阅读