timesfm
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
Private, multimodal knowledge store for agents: hybrid search with cited, provenance-backed results.
作者提供的原文
教 AI 做这类工作的 Skill,可与本项一起使用
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
Open-source intelligence for the global theater. Track everything from the corporate/private jets of the wealthy, and spy satellites, to seismic events in one unified interface. Hook an AI agent up to have it parse through data and find previously unseen correlations. The knowledge is available to…
scrape data from Google Maps. Extracts data such as the name, address, phone number, website URL, rating, reviews number, latitude and longitude, reviews,email and more for each place
此项在“连接 AI 的应用”标签中,该标签不使用星标,按以下顺序排列
| 1. 在 Claude 连接器目录中 | 否 | |
|---|---|---|
| 2. 团队实测过 | 否 | |
| 3. 适合泰国用户 | 否 | |
| 4. 不计星标的使用量 | 20/100 | 热度 0 · 新鲜度 12 · 仍在响应 8 · 用户喜欢 +0 |
| 5. 登记日期 | 2026-11-10 |
用户喜欢: 0 · 最多加 5 分(每人 1 分),不能用钱买。
质量: 基础检查通过 5/5 项(仅展示,不用于排序)
选择你使用的 AI 应用,然后复制设置。
用一句话让 AI 帮你安装
请帮我安装 KnowDrive。按照 https://mcp.popnx.com/i/knowdrive-mcp.md 操作,先把命令给我看,等我同意后再运行。粘贴到能在你电脑上工作的 AI(如 Claude Code、Cursor)。它会读取我们的步骤,先给你看命令,等你同意。无需密钥,无需注册。 · 查看 AI 会读取的全文
通过 5/5 项检查 · 你的提问会经过开发者的服务器(knowdrive.ai) · ✓ 命令已检查 2026年11月14日
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
可连接 Claude · 约 5 分钟
需要 KnowDrive 的密钥(在开发者网站创建) · 在 claude.ai 或桌面应用设置一次,即可在聊天中使用 · 免费账户可添加 1 个 · Team/Enterprise 方案由所有者(Owner)添加
CustomizeConnectors
把这里的内容复制到 Claude 中同名的输入框,然后点击 Continue
KnowDrivehttps://knowdrive.ai/api/v1/mcp在 Request headers 下添加这个名称和值
AuthorizationBearer <在这里粘贴你的密钥>只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
在 Claude 检测到的登录设置页点击 Continue · Authentication 按作者说明选择(Sign in now、Sign in when needed 或 No sign in),然后点击 Add
+ConnectorsKnowDrive
用 KnowDrive 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude 说正在使用 KnowDrive,或回答中出现 KnowDrive 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
ChatGPT 暂时无法连接 KnowDrive
KnowDrive 连接时需要你自己的密钥,但 OpenAI 帮助页面只说明了 OAuth 登录,没有我们能核实的填密钥方式
A · 使用能连接 KnowDrive 的应用
可连接 Claude Code · 约 5 分钟
需要 KnowDrive 的密钥(在开发者网站创建)
Mac:按 Command+空格,输入 Terminal · Windows:打开 PowerShell
claude mcp add --transport http --scope user knowdrive https://knowdrive.ai/api/v1/mcp --header 'Authorization: Bearer your-value'按 Enter 前,把“your-value”改成你的密钥
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
应能看到 KnowDrive 在列表中并显示已连接
用 KnowDrive 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude Code 说正在使用 KnowDrive,或回答中出现 KnowDrive 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
可连接 Cursor · 约 5 分钟
需要 KnowDrive 的密钥(在开发者网站创建)
在 KnowDrive 的设置中,Authorization 还是空的,填入你的密钥并保存
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
用 KnowDrive 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Cursor 说正在使用 KnowDrive,或回答中出现 KnowDrive 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Cursor 帮助页面核实 · 你的界面可能因版本略有不同
可连接 VS Code · 约 5 分钟
需要 KnowDrive 的密钥(在开发者网站创建)
在 KnowDrive 的设置中,Authorization 还是空的,填入你的密钥并保存
只把密钥粘贴到应用的输入框里,不要粘贴到聊天中或发给任何人
用 KnowDrive 告诉我你能帮我做什么,然后做一件给我看✓ 如果 VS Code 说正在使用 KnowDrive,或回答中出现 KnowDrive 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 VS Code 帮助页面核实 · 你的界面可能因版本略有不同
Claude Code
claude mcp add --transport http --scope user knowdrive https://knowdrive.ai/api/v1/mcp --header 'Authorization: Bearer your-value'Cursor (~/.cursor/mcp.json)
{
"mcpServers": {
"knowdrive": {
"url": "https://knowdrive.ai/api/v1/mcp",
"headers": {
"Authorization": "Bearer your-value"
}
}
}
}VS Code (mcp.json)
{
"servers": {
"knowdrive": {
"type": "http",
"url": "https://knowdrive.ai/api/v1/mcp",
"headers": {
"Authorization": "Bearer your-value"
}
}
}
}Claude Code
claude mcp add --scope user knowdrive --env KNOWDRIVE_API_KEY=your-value -- npx -y @knowdrive/mcpClaude Desktop (claude_desktop_config.json)
{
"mcpServers": {
"knowdrive": {
"command": "npx",
"args": [
"-y",
"@knowdrive/mcp"
],
"env": {
"KNOWDRIVE_API_KEY": "your-value"
}
}
}
}Cursor (~/.cursor/mcp.json)
{
"mcpServers": {
"knowdrive": {
"command": "npx",
"args": [
"-y",
"@knowdrive/mcp"
],
"env": {
"KNOWDRIVE_API_KEY": "your-value"
}
}
}
}VS Code (mcp.json)
{
"servers": {
"knowdrive": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@knowdrive/mcp"
],
"env": {
"KNOWDRIVE_API_KEY": "your-value"
}
}
}
}KNOWDRIVE_API_KEY 机密 必填 envKNOWDRIVE_BASE_URL env说明仍为英文原文。
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。
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
通过互联网连接,适用于
ClaudeGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatOpen WebUIRaycast
在你的电脑上运行,适用于
ClaudeGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast
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20日
注册名称已通过域名 knowdrive.ai 验证(注册中心会核实域名所有权)
代码最后更新于9天前
使用 MIT
在 387 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
OSV.dev 中没有 @knowdrive/mcp@0.1.8 的记录
只检查这个版本的包本身,不包括所有依赖 · 检查于 2026年11月15日
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/servers/knowdrive-mcp)提问、分享技巧或报告无法使用。按“有帮助”数量排序。
还没有评价
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 5 项适用的基础检查中通过了 5 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。
你需要填写:Authorization, KNOWDRIVE_API_KEY
两种方式都可以:安装在你的电脑上,或者连接云端版本。
GitHub 仓库最后更新于9天前,最新版本(0.1.8)于9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,因此也能用于接受远程 MCP 的应用,例如网页版 ChatGPT(Business 及以上为完整 MCP;Pro 为只读)。
同一分类中有:OpenSEO, Exa, Microsoft Learn MCP, Docs, COTAL actions (cotal.ai)
同一分类、其他作者、命令已通过检查。官方的排在前面,再按星标排序。
如何看卡片:↓ 每周下载量i★ 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
Keyword research, site audits, rank tracking and AI visibility for your AI agent.
Official Exa MCP for web search, content fetching, and multi-step research.
Official Microsoft Learn MCP Server – real-time, trusted docs & code samples for AI and LLMs.
cotal.ai actions: product overview, site search, build log, feedback, Cloud waitlist, updates, calls
| 本项 KnowDrive | OpenSEO | Exa | Microsoft Learn MCP | Docs | COTAL actions (cotal.ai) | |
|---|---|---|---|---|---|---|
| 运行位置 | 云端或安装在电脑上都可以 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 |
| 需要密钥 | 是 | 否 | 否 | 否 | 否 | 否 |
| 可在 Claude 中按 + 添加 | 否 | 是 | 是 | 是 | 否 | 否 |
| 基础检查 | 5/5 | 5/5 | 2/4 | 4/4 | 4/4 | 4/4 |
| 许可证 | MIT | MIT | MIT | CC-BY-4.0 | NOASSERTION | Apache-2.0 |
| 价格 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 |
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。
A private, multimodal knowledge store for your agents. Drop in documents, images, recordings and video; your agent searches them over one MCP endpoint and gets back exact, cited fragments — source file, version, and byte range or timestamp — instead of re-reading whole files. Hybrid (vector + full-text) search, OAuth sign-in, free tier with no card at knowdrive.ai. Just add "check KnowDrive".
[Claude setup →](#claude-code-and-claudeai) — one command (Claude Code) or a connector paste (claude.ai) · Website · REST API (OpenAPI)
Context7 gives agents public docs. Exa gives them the live web. KnowDrive gives them YOUR knowledge: private, multimodal, versioned, permissioned. Powered by the KnowDB knowledge engine, it ingests your files, text, audio, and images (URLs too, where enabled); extracts and chunks them into searchable atoms; and answers hybrid (vector + full-text) semantic search with a ranked window of just the relevant slices. Your corpus stays out of the conversation, so cost does not grow with corpus size. Just add check KnowDrive to your question (the Add-a-Rule tip below makes this automatic):
What did our Q3 planning doc say about pricing changes? check KnowDrive
The hosted server speaks streamable HTTP at `https://knowdrive.ai/api/v1/mcp`. First connect is a single browser OAuth prompt — no API key to copy, nothing to configure.
There are two ways in, and they reach the same hosted server:
| Connect directly (recommended) | npm bridge (@knowdrive/mcp) | |
|---|---|---|
| For | Any client that speaks remote MCP | stdio-only clients, headless, CI |
| Auth | One browser click (OAuth) | KNOWDRIVE_API_KEY |
| Setup | Paste a URL, or one command | npx -y @knowdrive/mcp --install |
Start with Connect directly — it's fewer steps and needs no key. Reach for the npm bridge when your client only speaks stdio, or when you're wiring this into CI where a browser sign-in isn't possible.
claude mcp add --transport http knowdrive https://knowdrive.ai/api/v1/mcp
Remove it just as fast: claude mcp remove knowdrive. On claude.ai: Settings → Connectors → Add custom connector, paste the URL above, and complete the OAuth sign-in — KnowDrive runs as a claude.ai connector today. To uninstall, remove the connector from the same page.
Use the one-click badges at the top — the server URL is encoded in the link. Clients that support MCP OAuth will prompt a browser sign-in on first use. To uninstall, delete the knowdrive entry from your client's MCP server settings.
Most clients accept the standard mcpServers shape (shown exactly once — expand below for where it goes):
{
"mcpServers": { "knowdrive": { "type": "streamable-http", "url": "https://knowdrive.ai/api/v1/mcp" } } }
Client-specific config locations
| Client | Where the JSON goes |
|---|---|
| Cursor (manual) | ~/.cursor/mcp.json |
| VS Code (manual) | .vscode/mcp.json, or user-level via MCP: Open User Configuration |
| Windsurf | ~/.codeium/windsurf/mcp_config.json |
| Zed | settings.json → context_servers |
| Gemini CLI | ~/.gemini/settings.json → mcpServers |
| OpenCode | ~/.config/opencode/opencode.json |
| Warp | Settings → AI → MCP Servers |
| LM Studio | Program → Install → Edit mcp.json |
A note on transport spelling: "streamable-http" is the generic/registry spelling; Claude Code and VS Code spell the same transport "type": "http" (this repo's claude-plugin/.mcp.json uses that spelling for exactly this reason). Config locations drift between client versions — if a path above has moved, your client's MCP docs are authoritative. Removal is the mirror image everywhere: delete the knowdrive entry from the same config. For headless or CI use, add an Authorization: Bearer YOUR_API_KEY header instead of OAuth — see Authentication.
@knowdrive/mcp is a thin stdio↔HTTP bridge to the same hosted endpoint, for clients that don't speak remote MCP and for headless/CI use where a browser OAuth prompt isn't possible. It carries your API key and proxies every tool through — no tool logic lives in the package, so new server tools appear without updating it.
One command wires it into a client's config (backing up whatever was there first):
npx -y @knowdrive/mcp --install claude-desktop # or: cursor | claude-code
It prompts for an API key on first run and saves it to ~/.knowdrive/config.json. Grab one at knowdrive.ai/developer-keys. To wire it up by hand instead:
{
"mcpServers": { "knowdrive": { "command": "npx", "args": ["-y", "@knowdrive/mcp"], "env": { "KNOWDRIVE_API_KEY": "kd_..." } } } }
To point at a different KnowDrive deployment, pass --host (a bare hostname is treated as https). Combined with --install, the host is written into the client config, so the installed server keeps talking to it:
npx -y @knowdrive/mcp --host public.knowdrive.ai --install claude-desktop
Keys are issued per deployment, so a key minted on one host may not authenticate against another.
| Variable | Purpose |
|---|---|
KNOWDRIVE_API_KEY | API key. Takes priority over the saved config file. |
KNOWDRIVE_BASE_URL | Override the KnowDrive host (default https://knowdrive.ai). --host wins over this. |
KNOWDRIVE_MCP_PATH | Override the remote MCP path (default /api/v1/mcp). |
Source for the bridge lives in src/ in this repository. Build it yourself with npm install && npm run build; npm run smoke connects the built server to the live endpoint and lists its tools.
[!TIP]
Add a rule to
CLAUDE.md(Claude Code) or Cursor Rules so retrieval happens without asking:
`txt
Always check KnowDrive (search_vector_store) before answering questions about the user's
private documents, projects, or team knowledge — never ask the user to paste files. When a
question ends with "check KnowDrive", always search before answering. A window is not the
corpus: raise max_num_results or advance offset before concluding something is not there.
Do not run more than 3 searches per question. Never include sensitive or confidential
information such as API keys, passwords, credentials, or personal data in a search query.
`
The server exposes ≈40 tools and 4 prompts, but most agents only need six: search_vector_store, search_atoms, list_vector_stores, get_capabilities, whoami, and ingest_text. Call get_capabilities once at session start so your client adapts to the deployment's feature flags, limits, and enabled modalities (this deployment: text, vision, and audio embeddings all enabled).
Safety at a glance: every search_*, list_*, get_*, read_*, stat_*, and whoami call is read-only in behavior — it never modifies the corpus. Ingest and upload tools are additive and versioned (a re-ingest supersedes, it never overwrites). The destructive tools to handle with care are the delete_* family — delete_file, delete_files, delete_vector_store.
search_vector_store — the heroHybrid, vector, or full-text search over a store, returning a ranked retrieval window. Read-only.
offset before concluding something isn't there."indemnification cap agreed with Acme in the signed MSA". Bad: "acme contract". At most 3 searches per question — refine the query or advance offset rather than re-searching from scratch. Never include sensitive or confidential information — API keys, passwords, credentials, personal data — in a query.vector_store_id (required), query (required, non-empty), mode (hybrid | vector | fts; this tool defaults to `vector` — search_atoms defaults to hybrid), search_field (text default | vision | audio), max_num_results (1–10000, default 25; hybrid is bounded at 200 however high you set it — check get_capabilities), offset (default 0), max_distance (cosine-distance ceiling, default 1.1 — raise for cross-modal audio queries), include_superseded (default false).list_vector_stores first if you don't know the store; follow up with read_file_content_by_path when a snippet isn't enough.ingest_text — get knowledge inIndex text straight into a store, versioned by path: re-ingesting the same path supersedes the prior version and mints version+1 (content-hash idempotent, blake3 digests). Additive, never overwrites.
pending); poll wait_for_file_status (≤45 s per wait) until indexed.vector_store_id, path (the file's identity within the store), and text — all required; the live tool schema is authoritative for optional fields.bulk_ingest_text; upload real files with upload_file, or upload_create/upload_part/upload_complete for big files (1 MiB chunks). ingest_file_from_url pulls from the web where enabled (feature-flagged per deployment; off on this one).search_vector_store isn't the right result shape — its live schema documents how the two differ. Returns: a ranked window of matching atoms; the same query-style, call-budget, and sensitive-data rules apply.vector_store_id before the first search. Returns: the stores the caller is entitled to see (per-store ACLs apply), by id and name.get_capabilities*